CN115357780A - Information recommendation method and device, computer equipment and storage medium - Google Patents

Abstract

The application relates to an information recommendation method, an information recommendation device, computer equipment and a storage medium. The method comprises the following steps: acquiring account interactive characteristics corresponding to a social network to which a target account belongs; the account interaction characteristics are obtained according to interaction behaviors among all social accounts in the social network; performing characteristic propagation mapping on account characteristics corresponding to the target account through account interactive characteristics, and obtaining propagation account characteristics corresponding to the target account according to the result of the characteristic propagation mapping; acquiring respective information interaction characteristics of each piece of information to be recommended; the information interaction characteristics are generated based on account characteristics of social accounts in the social network, which are subjected to interaction operation on corresponding information to be recommended; and matching the propagation account characteristics with the information interaction characteristics, and recommending the target information determined from the information to be recommended to the terminal corresponding to the target account according to the matching result. By adopting the method, the accuracy of information recommendation can be improved.

Description

Translated fromChinese

信息推荐方法、装置、计算机设备和存储介质Information recommendation method, device, computer equipment and storage medium

技术领域technical field

本申请涉及计算机技术领域，特别是涉及一种信息推荐方法、装置、计算机设备和存储介质。The present application relates to the field of computer technology, in particular to an information recommendation method, device, computer equipment and storage medium.

背景技术Background technique

随着计算机和互联网技术的快速发展，为人们的生活带来了许多便利，同时也带了海量的数据信息，使得人们难以从大量的数据信息中获取所需的信息。例如，对于互联网中海量的音视频、文章、图片、网页等各种数据信息，人们很难快速过滤获得所需要的信息。为了让用户更准确地获取到所需要的信息，往往通过个性化信息推荐，以为用户推荐满足用户需要的信息。With the rapid development of computer and Internet technology, it has brought a lot of convenience to people's life, but also brought massive data information, making it difficult for people to obtain the required information from a large amount of data information. For example, it is difficult for people to quickly filter and obtain the required information for a large amount of data information such as audio and video, articles, pictures, and web pages on the Internet. In order to allow users to obtain the required information more accurately, personalized information recommendation is often used to recommend information that meets the user's needs.

目前，信息推荐主要是对用户的历史行为进行分析，如对用户的视频浏览行为进行分析，根据分析得到的用户偏好推荐相应的信息。然而，在用户的历史行为数据较少时，对历史行为的分析无法有效获得用户偏好，导致针对用户的信息推荐准确性有限。At present, information recommendation is mainly to analyze the user's historical behavior, such as analyzing the user's video browsing behavior, and recommend corresponding information according to the user preference obtained from the analysis. However, when the user's historical behavior data is small, the analysis of historical behavior cannot effectively obtain user preferences, resulting in limited accuracy of information recommendation for users.

发明内容Contents of the invention

基于此，有必要针对上述技术问题，提供一种能够提高信息推荐准确性的信息推荐方法、装置、计算机设备和存储介质。Based on this, it is necessary to provide an information recommendation method, device, computer equipment and storage medium capable of improving the accuracy of information recommendation for the above technical problems.

一种信息推荐方法，所述方法包括：A method for recommending information, the method comprising:

获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到；Acquiring the account interaction features corresponding to the social network to which the target account belongs; the account interaction features are obtained according to the interaction behavior between social accounts in the social network;

通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征；Perform feature propagation mapping on the account features corresponding to the target account through the account interaction features, and obtain the propagating account features corresponding to the target account according to the result of the feature propagation mapping;

获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；Obtain the information interaction characteristics of each information to be recommended; the information interaction characteristics are generated based on the account characteristics of social accounts that have interacted with the corresponding information to be recommended in the social network;

将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。The characteristics of the dissemination account are matched with the characteristics of each information interaction, and the target information determined from each information to be recommended is recommended to the terminal corresponding to the target account according to the matching result.

一种信息推荐装置，所述装置包括：An information recommendation device, the device comprising:

账号交互特征获取模块，用于获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到；The account interaction feature acquisition module is used to acquire the account interaction feature corresponding to the social network to which the target account belongs; the account interaction feature is obtained according to the interaction behavior between social accounts in the social network;

特征传播映射模块，用于通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征；The feature propagation mapping module is used to perform feature propagation mapping on the account characteristics corresponding to the target account through the account interaction characteristics, and obtain the propagation account characteristics corresponding to the target account according to the result of the feature propagation mapping;

信息交互特征获取模块，用于获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；The information interaction feature acquisition module is used to acquire the respective information interaction features of each information to be recommended; the information interaction features are generated based on the account features of social accounts that have interacted with the corresponding information to be recommended in the social network;

目标信息推荐模块，用于将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。The target information recommendation module is used to match the characteristics of the dissemination account and the interaction features of each information, and recommend the target information determined from each information to be recommended according to the matching result to the terminal corresponding to the target account.

在其中一个实施例中，特征传播映射模块包括：In one of the embodiments, the feature propagation mapping module includes:

目标交互特征提取模块，用于从账号交互特征中提取与目标账号对应的目标账号交互特征；The target interaction feature extraction module is used to extract the target account interaction feature corresponding to the target account from the account interaction feature;

交互账号特征获取模块，用于获取目标账号交互特征对应的交互社交账号的账号特征；An interactive account feature acquisition module, configured to acquire account features of an interactive social account corresponding to the target account interaction feature;

传播映射迭代模块，用于基于目标账号交互特征、交互社交账号的账号特征和传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果；The propagation mapping iteration module is used to perform iterative feature propagation mapping on the account characteristics corresponding to the target account based on the interaction characteristics of the target account, the account characteristics of the interactive social account and the propagation mapping parameters, and obtain the result of the characteristic propagation mapping;

传播账号特征确定模块，用于根据特征传播映射的结果确定目标账号对应的传播账号特征。The dissemination account feature determination module is configured to determine the dissemination account features corresponding to the target account according to the result of the feature dissemination mapping.

在其中一个实施例中，传播映射迭代模块包括：In one of the embodiments, the propagation map iteration module includes:

当前账号特征确定模块，用于将目标账号对应的账号特征确定为当前账号特征；The current account feature determination module is used to determine the account feature corresponding to the target account as the current account feature;

传播映射处理模块，用于基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果；将本次迭代特征传播映射的结果作为当前账号特征，并返回基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果的步骤。The propagation mapping processing module is used to perform feature propagation mapping on the characteristics of the current account based on the interaction characteristics of the target account, the account characteristics of the interactive social account and the propagation mapping parameters of this iterative feature propagation mapping, and obtain the result of this iterative feature propagation mapping; Use the result of this iterative feature propagation mapping as the current account feature, and return the propagation mapping parameters based on the target account interaction feature, the account feature of the interactive social account, and this iterative feature propagation mapping, and perform feature propagation mapping on the current account feature to get This step to iterate over the results of the feature propagation map.

在其中一个实施例中，传播映射处理模块包括：In one of the embodiments, the propagation mapping processing module includes:

特征传播模块，用于基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播，得到特征传播结果；The feature propagation module is used to perform feature propagation on the current account feature based on the interaction feature of the target account, the account feature of the interactive social account, and the propagation mapping parameters of this iterative feature propagation map, and obtain the feature propagation result;

非线性映射模块，用于对特征传播结果进行非线性映射，得到本次迭代特征传播映射的结果。The nonlinear mapping module is used to perform nonlinear mapping on the feature propagation result to obtain the result of this iterative feature propagation mapping.

在其中一个实施例中，特征传播映射模块包括：In one of the embodiments, the feature propagation mapping module includes:

网络账号特征确定模块，用于确定根据社交网络中各社交账号的账号特征构建的网络账号特征；网络账号特征包括目标账号对应的账号特征；The network account feature determination module is used to determine the network account features constructed according to the account features of each social account in the social network; the network account features include account features corresponding to the target account;

模型特征传播映射模块，用于将账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射，获得特征传播映射模型输出的网络传播账号特征；The model feature propagation mapping module is used to input the account interaction features and network account features into the feature propagation mapping model to perform feature propagation mapping, and obtain the network propagation account characteristics output by the feature propagation mapping model;

模型输出处理模块，用于从网络传播账号特征中提取目标账号对应的传播账号特征。The model output processing module is used to extract the characteristics of the communication account corresponding to the target account from the characteristics of the network communication account.

在其中一个实施例中，模型特征传播映射模块包括：In one of the embodiments, the model feature propagation mapping module includes:

标准化处理模块，用于通过标准化条件对账号交互特征进行标准化处理，得到标准化的账号交互特征；A standardized processing module, configured to perform standardized processing on account interaction features through standardized conditions to obtain standardized account interaction features;

特征输入模块，用于将标准化的账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射。The feature input module is used to input standardized account interaction features and network account features into the feature propagation mapping model for feature propagation mapping.

在其中一个实施例中，目标信息推荐模块包括：In one of the embodiments, the target information recommendation module includes:

特征匹配模块，用于通过匹配模型对传播账号特征和各信息交互特征进行匹配，得到匹配模型输出的匹配结果；The feature matching module is used to match the characteristics of the dissemination account and each information interaction feature through the matching model, and obtain the matching result output by the matching model;

目标信息确定模块，用于基于匹配结果从各待推荐信息中确定目标信息；A target information determination module, configured to determine the target information from each information to be recommended based on the matching result;

目标信息处理模块，用于将目标信息向目标账号对应的终端进行推荐。The target information processing module is configured to recommend the target information to the terminal corresponding to the target account.

在其中一个实施例中，目标信息确定模块包括：In one of the embodiments, the target information determination module includes:

推荐条件筛选模块，用于从匹配结果中确定满足推荐条件的目标匹配结果；The recommendation condition screening module is used to determine the target matching result satisfying the recommendation condition from the matching results;

筛选结果处理模块，用于将目标匹配结果所对应的待推荐信息确定为目标信息。The screening result processing module is configured to determine the information to be recommended corresponding to the target matching result as the target information.

在其中一个实施例中，所述装置还包括：In one of the embodiments, the device also includes:

统计结果确定模块，用于确定目标账号所属的社交网络中各社交账号之间交互行为的统计结果；A statistical result determination module, configured to determine statistical results of interaction behaviors between social accounts in the social network to which the target account belongs;

交互特征获得模块，用于基于统计结果获得社交网络中各社交账号之间的交互特征；An interactive feature obtaining module, configured to obtain interactive features between social accounts in the social network based on statistical results;

账号交互特征生成模块，用于根据社交网络中各社交账号之间的交互特征，生成社交网络对应的账号交互特征。The account interaction feature generation module is configured to generate account interaction features corresponding to the social network according to the interaction features between social accounts in the social network.

在其中一个实施例中，所述装置还包括：In one of the embodiments, the device also includes:

社交网络确定模块，用于确定目标账号所属的社交网络；A social network determining module, configured to determine the social network to which the target account belongs;

节点嵌入模块，用于对社交网络中的各节点进行节点嵌入，得到各节点分别对应的节点特征；各节点与社交网络中的各社交账号对应，各节点之间的节点关系与各社交账号之间的交互行为对应；The node embedding module is used to embed each node in the social network to obtain the node characteristics corresponding to each node; each node corresponds to each social account in the social network, and the node relationship between each node is related to the relationship between each social account. Correspondence between interactive behaviors;

账号特征确定模块，用于基于各节点对应的节点特征，确定目标账号对应的账号特征。The account feature determining module is configured to determine the account feature corresponding to the target account based on the node features corresponding to each node.

在其中一个实施例中，节点嵌入模块：In one of these embodiments, the node embedding module:

权值确定模块，用于根据社交网络中各节点之间的节点关系，确定各节点之间的游走权重；The weight determination module is used to determine the walking weight between each node according to the node relationship between each node in the social network;

游走模块，用于以每个节点为起点，基于游走权重在社交网络中进行节点游走，形成各节点游走轨迹；The walking module is used to take each node as the starting point and perform node walking in the social network based on the walking weight to form the walking trajectory of each node;

游走轨迹处理模块，用于通过嵌入模型对各节点游走轨迹进行特征嵌入，得到各节点分别对应的节点特征。The walking trajectory processing module is used to perform feature embedding on the walking trajectory of each node through the embedding model, and obtain the node characteristics corresponding to each node.

在其中一个实施例中，所述装置还包括：In one of the embodiments, the device also includes:

交互社交账号确定模块，用于确定每个待推荐信息对各自的交互社交账号；交互社交账号为社交网络中对相应待推荐信息产生过交互操作的社交账号；The interactive social account determination module is used to determine the respective interactive social account of each information to be recommended; the interactive social account is a social account that has interacted with the corresponding information to be recommended in the social network;

信息交互特征获得模块，用于对各交互社交账号对应的账号特征进行特征聚合，得到相应待推荐信息的信息交互特征。The information interaction feature acquisition module is used to perform feature aggregation on the account features corresponding to each interactive social account, and obtain the information interaction features of the corresponding information to be recommended.

一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现以下步骤：A computer device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到；Acquiring the account interaction features corresponding to the social network to which the target account belongs; the account interaction features are obtained according to the interaction behavior between social accounts in the social network;

通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征；Perform feature propagation mapping on the account features corresponding to the target account through the account interaction features, and obtain the propagating account features corresponding to the target account according to the result of the feature propagation mapping;

获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；Obtain the information interaction characteristics of each information to be recommended; the information interaction characteristics are generated based on the account characteristics of social accounts that have interacted with the corresponding information to be recommended in the social network;

将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。The characteristics of the dissemination account are matched with the characteristics of each information interaction, and the target information determined from each information to be recommended is recommended to the terminal corresponding to the target account according to the matching result.

一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现以下步骤：A computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到；Acquiring the account interaction features corresponding to the social network to which the target account belongs; the account interaction features are obtained according to the interaction behavior between social accounts in the social network;

通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征；Perform feature propagation mapping on the account features corresponding to the target account through the account interaction features, and obtain the propagating account features corresponding to the target account according to the result of the feature propagation mapping;

获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；Obtain the information interaction characteristics of each information to be recommended; the information interaction characteristics are generated based on the account characteristics of social accounts that have interacted with the corresponding information to be recommended in the social network;

将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。The characteristics of the dissemination account are matched with the characteristics of each information interaction, and the target information determined from each information to be recommended is recommended to the terminal corresponding to the target account according to the matching result.

上述信息推荐方法、装置、计算机设备和存储介质，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征，并将传播账号特征与每个待推荐信息各自的信息交互特征进行匹配，信息交互特征基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成，根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。在信息推荐处理过程中，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，从而将社交网络中与目标账号具有交互行为关系的社交账号的账号特征传播映射至目标账号，得到目标账号对应的传播账号特征，并通过传播账号特征和待推荐信息各自的信息交互特征进行匹配，根据匹配结果确定目标信息进行推荐，从而利用了社交网络中各社交账号之间的交互行为关系进行信息推荐，提高了信息推荐的准确性。The above-mentioned information recommendation method, device, computer equipment and storage medium perform feature propagation mapping on the account features corresponding to the target account through the account interaction features obtained according to the interaction behavior between social accounts in the social network, and perform feature propagation mapping according to the result of the feature propagation mapping Obtain the characteristics of the communication account corresponding to the target account, and match the characteristics of the communication account with the information interaction characteristics of each information to be recommended. The information interaction characteristics are based on the account characteristics of the social accounts that have interacted with the corresponding information to be recommended in the social network. Generate, and recommend to the terminal corresponding to the target account from the target information determined in each information to be recommended according to the matching result. In the process of information recommendation processing, through the account interaction features obtained according to the interaction behavior between social accounts in the social network, the account features corresponding to the target account are subjected to feature propagation mapping, so that the social network has an interactive behavior relationship with the target account. The account features of the social account are propagated and mapped to the target account, and the corresponding spread account features of the target account are obtained, and the spread account features are matched with the information interaction features of the information to be recommended, and the target information is determined according to the matching results for recommendation, thus using Information recommendation is performed based on the interactive behavior relationship between social accounts in the social network, which improves the accuracy of information recommendation.

附图说明Description of drawings

图1为一个实施例中信息推荐方法的应用环境图；Fig. 1 is an application environment diagram of an information recommendation method in an embodiment;

图2为一个实施例中信息推荐方法的流程示意图；Fig. 2 is a schematic flow chart of an information recommendation method in an embodiment;

图3为一个实施例中特征传播映射的流程示意图；Fig. 3 is a schematic flow chart of feature propagation mapping in an embodiment;

图4为一个实施例中社交网络中社交账号之间交互行为的示意图；FIG. 4 is a schematic diagram of interaction between social accounts in a social network in an embodiment;

图5为另一个实施例中特征传播映射的流程示意图；Fig. 5 is a schematic flow chart of feature propagation mapping in another embodiment;

图6为一个实施例中推荐目标信息的流程示意图；Fig. 6 is a schematic flow chart of recommending target information in an embodiment;

图7为另一个实施例中社交网络中社交账号之间交互行为的示意图；FIG. 7 is a schematic diagram of interaction between social accounts in a social network in another embodiment;

图8为图7所示实施例中社交网络对应节点结构的示意图；Fig. 8 is a schematic diagram of the corresponding node structure of the social network in the embodiment shown in Fig. 7;

图9为另一个实施例中信息推荐方法的流程示意图；FIG. 9 is a schematic flowchart of an information recommendation method in another embodiment;

图10为一个实施例中信息推荐装置的结构框图；Fig. 10 is a structural block diagram of an information recommendation device in an embodiment;

图11为一个实施例中计算机设备的内部结构图。Figure 11 is a diagram of the internal structure of a computer device in one embodiment.

具体实施方式Detailed ways

为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not intended to limit the present application.

本申请提供的信息推荐方法，可以应用于如图1所示的应用环境中。其中，终端102通过网络与服务器104进行通信。用户A、B、C和D在同一社交网络中具有社交账号，用户C与用户A、B和D在社交网络中均产生过交互行为。用户C在终端102通过社交网络的目标账号登录短视频平台，在用户C点击短视频推荐控件，触发进行短视频推荐时，服务器104响应于终端102发送的短视频推荐请求，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征，并将传播账号特征与每个待推荐短视频各自的信息交互特征进行匹配，信息交互特征基于社交网络中对相应待推荐短视频产生过交互操作的社交账号的账号特征生成，根据匹配结果从各待推荐短视频中确定的目标短视频向目标账号对应的终端102进行推荐，终端102接收到服务器104推荐的目标短视频后进行展示。此外，信息推荐所针对推荐信息可以不限于短视频，还可以包括音视频、图片、文本、网页、名片等各种数据信息。The information recommendation method provided in this application can be applied to the application environment shown in FIG. 1 . Wherein, the terminal 102 communicates with theserver 104 through the network. Users A, B, C and D have social accounts in the same social network, and user C has interacted with users A, B and D in the social network. User C logs in to the short video platform through the target account of the social network on theterminal 102. When user C clicks on the short video recommendation control to trigger the short video recommendation, theserver 104 responds to the short video recommendation request sent by the terminal 102 by Based on the account interaction features obtained from the interaction behaviors between social accounts, feature propagation mapping is performed on the account features corresponding to the target account, and the propagation account characteristics corresponding to the target account are obtained according to the result of feature propagation mapping, and the propagation account characteristics are compared with each waiting account. The information interaction characteristics of the recommended short videos are matched. The information interaction characteristics are generated based on the account characteristics of the social accounts that have interacted with the corresponding short videos to be recommended in the social network. According to the matching results, the target short videos determined from each short video to be recommended The video is recommended to the terminal 102 corresponding to the target account, and the terminal 102 displays the target short video after receiving the target short video recommended by theserver 104 . In addition, the recommended information for information recommendation may not be limited to short videos, but may also include various data information such as audio and video, pictures, texts, web pages, and business cards.

其中，终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑、车载设备和便携式可穿戴设备，服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现，服务器104还可以是提供云服务、云数据库、云计算、云函数、云存储、网络服务、云通信、中间件服务、域名服务、安全服务、CDN、以及大数据和人工智能平台等基础云计算服务的云服务器。Wherein, the terminal 102 can be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, vehicle-mounted devices and portable wearable devices, and theserver 104 can be realized by an independent server or a server cluster composed of multiple servers, Theserver 104 can also provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. Cloud server for the service.

在具体应用中，信息推荐中涉及的账号交互特征、信息交互特征等数据，还可以上链至区块链中进行安全保存。区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层。In specific applications, data such as account interaction characteristics and information interaction characteristics involved in information recommendation can also be uploaded to the blockchain for safe storage. Blockchain is a new application model of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain (Blockchain), essentially a decentralized database, is a series of data blocks associated with each other using cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

区块链底层平台可以包括用户管理、基础服务、智能合约以及运营监控等处理模块。其中，用户管理模块负责所有区块链参与者的身份信息管理，包括维护公私钥生成(账户管理)、密钥管理以及用户真实身份和区块链地址对应关系维护(权限管理)等，并且在授权的情况下，监管和审计某些真实身份的交易情况，提供风险控制的规则配置(风控审计)；基础服务模块部署在所有区块链节点设备上，用来验证业务请求的有效性，并对有效请求完成共识后记录到存储上，对于一个新的业务请求，基础服务先对接口适配解析和鉴权处理(接口适配)，然后通过共识算法将业务信息加密(共识管理)，在加密之后完整一致的传输至共享账本上(网络通信)，并进行记录存储；智能合约模块负责合约的注册发行以及合约触发和合约执行，开发人员可以通过某种编程语言定义合约逻辑，发布到区块链上(合约注册)，根据合约条款的逻辑，调用密钥或者其它的事件触发执行，完成合约逻辑，同时还提供对合约升级注销的功能；运营监控模块主要负责产品发布过程中的部署、配置的修改、合约设置、云适配以及产品运行中的实时状态的可视化输出，例如：告警、监控网络情况、监控节点设备健康状态等。The underlying blockchain platform can include processing modules such as user management, basic services, smart contracts, and operational monitoring. Among them, the user management module is responsible for the identity information management of all blockchain participants, including maintenance of public and private key generation (account management), key management, and maintenance of the corresponding relationship between the user's real identity and blockchain address (authority management), etc., and in In the case of authorization, supervise and audit transactions of certain real identities, and provide risk control rule configuration (risk control audit); the basic service module is deployed on all blockchain node devices to verify the validity of business requests, And complete the consensus on valid requests and record them on the storage. For a new business request, the basic service first analyzes and authenticates the interface adaptation (interface adaptation), and then encrypts the business information through the consensus algorithm (consensus management). After encryption, it is completely and consistently transmitted to the shared ledger (network communication) and recorded for storage; the smart contract module is responsible for the registration and issuance of the contract, contract triggering and contract execution. Developers can define the contract logic through a programming language and publish it to On the blockchain (contract registration), according to the logic of the contract terms, call the key or other events to trigger execution, complete the contract logic, and also provide the function of contract upgrade and cancellation; the operation monitoring module is mainly responsible for the deployment during the product release process , configuration modification, contract setting, cloud adaptation, and visual output of real-time status during product operation, such as: alarms, monitoring network conditions, monitoring node device health status, etc.

平台产品服务层提供典型应用的基本能力和实现框架，开发人员可以基于这些基本能力，叠加业务的特性，完成业务逻辑的区块链实现。应用服务层提供基于区块链方案的应用服务给业务参与方进行使用。The platform product service layer provides the basic capabilities and implementation framework of typical applications. Based on these basic capabilities, developers can superimpose the characteristics of the business and complete the blockchain implementation of business logic. The application service layer provides application services based on blockchain solutions for business participants to use.

在一个实施例中，如图2所示，提供了一种信息推荐方法，以该方法应用于图1中的服务器为例进行说明，包括以下步骤：In one embodiment, as shown in FIG. 2 , a method for recommending information is provided. The method is applied to the server in FIG. 1 as an example for illustration, including the following steps:

步骤202，获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到。Instep 202, the account interaction features corresponding to the social network to which the target account belongs are obtained; the account interaction features are obtained according to the interaction behavior between social accounts in the social network.

其中，目标账号为需要进行信息推荐的用户账号，目标账号具体可以为社交网络中的社交账号，如用户可以通过社交账号登录信息客户端，从而服务器可以通过用户登录的社交账号在相应社交网络中的社交关系，对用户进行准确的信息推荐。社交关系指社会上人与人通过社交网络的交际往来，是人们运用一定的方式传递信息、交流思想的意识，以达到某种目的的社会各项活动，具体可以为人们通过社交网络注册不同的社交账号进行交互，如聊天、点赞、评论等各种交互行为。Wherein, the target account is a user account that needs to perform information recommendation, and the target account may specifically be a social account in a social network. For example, a user can log in to the information client through a social account, so that the server can use the social account logged in by the user to register in the corresponding social network. social relationships, and provide accurate information recommendations to users. Social relationship refers to the communication between people in society through social networks. It is the consciousness of people to use certain methods to transmit information and exchange ideas in order to achieve a certain purpose. Specifically, people can register different social networks through social networks. Interact with social accounts, such as chatting, likes, comments and other interactive behaviors.

账号交互特征根据社交网络中各社交账号之间的交互行为得到，交互行为包括用户通过各社交账号进行的聊天、评论、点赞、转发等各种交互操作。账号交互特征具体可以根据对社交网络中各社交账号之间的交互行为进行交互分析获得的交互分析结果得到，如可以对社交网络中两两社交账号之间的交互行为分别进行统计，得到两两社交账号之间交互行为统计结果，基于交互行为统计结果构建两两社交账号之间的交互特征，并根据社交网络中各社交账号之间的交互特征，获得社交网络对应的账号交互特征。账号交互特征反映了社交网络中各社交账号支架的交互关系。The account interaction feature is obtained according to the interaction behavior between social accounts in the social network, and the interaction behavior includes chatting, commenting, liking, forwarding and other interactive operations performed by the user through each social account. Specifically, the account interaction characteristics can be obtained according to the interaction analysis results obtained by interactive analysis of the interaction behaviors between social accounts in the social network. According to the statistical results of interaction behavior between social accounts, the interaction characteristics between two social accounts are constructed based on the interaction behavior statistics results, and the corresponding account interaction characteristics of the social network are obtained according to the interaction characteristics between social accounts in the social network. The account interaction features reflect the interaction relationship of each social account bracket in the social network.

在具体应用中，如对于社交网络中包括社交账号A、B、C和D，则可以分别对两两社交账号之间的交互行为进行分析，获得两两社交账号之间交互行为的统计结果，基于该统计结果可以进行特征提取，得到两两社交账号之间的交互特征，如可以根据两两社交账号之间聊天的次数、频率、交互累计时长等统计结果，综合确定两两社交账号之间的交互特征。获得社交网络中两两社交账号之间的交互特征之后，将各交互特征进行融合，得到社交网络对应的账号交互特征，账号交互特征可以携带有社交网络中两两账号之间的交互特征。In a specific application, if the social network includes social accounts A, B, C, and D, the interaction behavior between the two social accounts can be analyzed respectively, and the statistical results of the interaction behavior between the two social accounts can be obtained. Based on the statistical results, feature extraction can be carried out to obtain the interaction characteristics between two social accounts. For example, according to the statistical results such as the number of chats between two social accounts, the frequency, and the cumulative duration of interaction, etc., the interaction between two social accounts can be comprehensively determined. interactive features. After the interaction features between two social accounts in the social network are obtained, the interaction features are fused to obtain the account interaction features corresponding to the social network, and the account interaction features may carry the interaction features between the two accounts in the social network.

具体地，在触发针对目标账号进行信息推荐处理时，如服务器接收到终端上传的信息推荐请求时，服务器响应于该信息推荐请求，获取目标账号所属社交网络对应的账号交互特征，账号交互特征可以由服务器预先根据社交网络中的各社交账号之间的交互行为确定。Specifically, when triggering information recommendation processing for a target account, for example, when the server receives an information recommendation request uploaded by a terminal, the server responds to the information recommendation request by obtaining account interaction features corresponding to the social network to which the target account belongs. The account interaction features can be It is determined by the server in advance according to the interaction behavior between social accounts in the social network.

步骤204，通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征。Step 204, perform feature propagation mapping on account features corresponding to the target account through account interaction features, and obtain propagation account features corresponding to the target account according to the result of feature propagation mapping.

其中，目标账号对应的账号特征可以为表征目标账号本质特性的信息，具体可以通过对目标账号进行特征工程处理得到，特征工程是将原始数据转化成更好的表达问题本质的特征的过程，使得将这些特征运用到预测模型中能提高对不可见数据的模型预测精度。特征传播映射为利用社交网络中与目标账号存在社交关系的社交账号的特征描述目标账号特征的过程。对目标账号对应的账号特征进行特征传播映射，可以将与目标账号存在社交关系的社交账号的特征传递至目标账号，从而通过与目标账号存在社交关系的社交账号的特征对目标账号的账号特征进行进一步描述，从而获得能够准确反映目标账号社交关系的账号特征。目标账号对应的传播账号特征为目标账号对应的账号特征经过特征传播映射处理后获得的特征，传播账号特征除携带目标账号本身的特征外，还携带有社交网络中与目标账号存在社交关系的社交账号的账号特征。Among them, the account characteristics corresponding to the target account can be information representing the essential characteristics of the target account, which can be obtained by performing feature engineering processing on the target account. Feature engineering is a process of transforming raw data into features that better express the nature of the problem, so that Applying these features to predictive models can improve model prediction accuracy for unseen data. Feature propagation mapping is the process of describing the features of the target account by using the features of the social accounts that have a social relationship with the target account in the social network. By performing feature propagation mapping on the account features corresponding to the target account, the features of the social accounts that have a social relationship with the target account can be transferred to the target account, so that the account features of the target account can be analyzed by the features of the social accounts that have a social relationship with the target account. Further description, so as to obtain the account characteristics that can accurately reflect the social relationship of the target account. The propagation account features corresponding to the target account are the features obtained after the feature propagation mapping processing of the account features corresponding to the target account. In addition to the characteristics of the target account itself, the propagation account features also carry the social network information that has a social relationship with the target account in the social network. The account characteristics of the account.

具体地，服务器获得目标账号所属社交网络对应的账号交互特征后，进一步获得目标账号的账号特征，并通过该账号交互特征对目标账号的账号特征进行特征传播映射处理，得到特征传播映射的结果，服务器从特征传播映射的结果中获得目标账号对应的传播账号特征，传播账号特征结合了目标账号本身的账号特征以及在社交网络中存在社交关系的社交账号的账号特征。Specifically, after obtaining the account interaction characteristics corresponding to the social network to which the target account belongs, the server further obtains the account characteristics of the target account, and performs feature propagation mapping processing on the account characteristics of the target account through the account interaction characteristics, and obtains the result of feature propagation mapping, The server obtains the propagation account features corresponding to the target account from the result of the feature propagation mapping, and the propagation account features combine account features of the target account itself and account features of social accounts with social relationships in the social network.

步骤206，获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的。In step 206, the information interaction features of each information to be recommended are acquired; the information interaction features are generated based on the account features of social accounts in the social network that have interacted with the corresponding information to be recommended.

其中，待推荐信息为能够进行推荐的信息，具体可以为各种类型的信息，如音视频、图片、文本、网页、名片等各种数据信息。待推荐信息可以根据信息推荐实际所应用的场景进行确定，如对于短视频平台应用，待推荐信息可以为短视频平台应用中的各段短视频，以实现对目标账号准确推荐短视频。信息交互特征为每个待推荐信息的特征，信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的。在具体实现时，交互操作可以根据实际需要进行灵活设置，如可以设置为浏览、点赞、评论、转发、收藏、关注等各种操作，还可以为各种操作的组合。例如，对于短视频R1，在社交网络中的社交账号A、B和C均浏览过或点赞过该短视频R1，则可以认为社交账号A、B和C均对该短视频R1产生过交互操作，则可以根据社交账号A、B和C分别对应的账号特征，得到表征该短视频R1特性的信息交互特征。Wherein, the information to be recommended is information that can be recommended, and specifically may be various types of information, such as various data information such as audio and video, pictures, texts, web pages, and business cards. The information to be recommended can be determined according to the actual application scenario of the information recommendation. For example, for a short video platform application, the information to be recommended can be each short video in the short video platform application, so as to accurately recommend short videos to the target account. The information interaction feature is a feature of each information to be recommended, and the information interaction feature is generated based on account features of social accounts that have interacted with the corresponding information to be recommended in the social network. In actual implementation, interactive operations can be flexibly set according to actual needs, such as browsing, liking, commenting, reposting, favorites, following, etc., or a combination of various operations. For example, for a short video R1, if the social accounts A, B, and C in the social network have all browsed or liked the short video R1, then it can be considered that the social accounts A, B, and C have all interacted with the short video R1 operation, the information interaction features representing the characteristics of the short video R1 can be obtained according to the account features corresponding to the social accounts A, B, and C respectively.

具体地，服务器获取每个待推荐信息各自的信息交互特征，每个待推荐信息各自的信息交互特征可以由服务器预先基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成，以在触发信息推荐时获取各待推荐信息分别对应的信息交互特征。Specifically, the server acquires the information interaction characteristics of each information to be recommended, and the information interaction characteristics of each information to be recommended can be generated by the server in advance based on the account characteristics of social accounts that have interacted with the corresponding information to be recommended in the social network. , so as to obtain the information interaction features corresponding to each information to be recommended when information recommendation is triggered.

步骤208，将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。Step 208, matching the characteristics of the dissemination account with the interaction characteristics of each information, and recommending the target information determined from each information to be recommended according to the matching result to the terminal corresponding to the target account.

其中，目标信息为根据匹配结果从各待推荐信息中确定的需要向目标账号进行推荐的信息，即目标信息为从各待推荐信息中筛选得到的向目标账号进行推荐的信息。具体地，在得到目标账号的传播账号特征和待推荐信息分别对应的信息交互特征后，服务器将传播账号特征和各信息交互特征分别进行匹配，得到传播账号特征与各信息交互特征的匹配结果，并根据匹配结果从各待推荐信息中确定目标信息，服务器将目标信息向目标账号对应的终端进行推荐。Wherein, the target information is the information that needs to be recommended to the target account determined from the information to be recommended according to the matching result, that is, the target information is the information to be recommended to the target account that is screened from the information to be recommended. Specifically, after obtaining the dissemination account characteristics of the target account and the information interaction characteristics corresponding to the information to be recommended, the server respectively matches the dissemination account characteristics and each information interaction characteristic, and obtains the matching result of the dissemination account characteristics and each information interaction characteristic, And according to the matching result, the target information is determined from the information to be recommended, and the server recommends the target information to the terminal corresponding to the target account.

在具体实现时，可以分别确定传播账号特征和各信息交互特征的相似度，将相似度大于相似度阈值的信息交互特征所对应的待推荐信息，确定为目标信息，并将该目标信息向目标账号对应的终端进行推荐。In the actual implementation, the similarity between the dissemination account features and the information interaction features can be determined respectively, and the information to be recommended corresponding to the information interaction features whose similarity is greater than the similarity threshold is determined as the target information, and the target information is sent to the target The terminal corresponding to the account is recommended.

上述信息推荐方法中，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征，并将传播账号特征与每个待推荐信息各自的信息交互特征进行匹配，信息交互特征基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成，根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。在信息推荐处理过程中，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，从而将社交网络中与目标账号具有交互行为关系的社交账号的账号特征传播映射至目标账号，得到目标账号对应的传播账号特征，并通过传播账号特征和待推荐信息各自的信息交互特征进行匹配，根据匹配结果确定目标信息进行推荐，从而利用了社交网络中各社交账号之间的交互行为关系进行信息推荐，提高了信息推荐的准确性。In the above information recommendation method, the account feature corresponding to the target account is subjected to feature propagation mapping through the account interaction features obtained according to the interaction behavior between social accounts in the social network, and the propagation account corresponding to the target account is obtained according to the result of feature propagation mapping. characteristics, and match the dissemination account characteristics with the information interaction characteristics of each information to be recommended. The information interaction characteristics are generated based on the account characteristics of the social accounts that have interacted with the corresponding information to be recommended in the social network. The target information determined in the information to be recommended is recommended to the terminal corresponding to the target account. In the process of information recommendation processing, through the account interaction features obtained according to the interaction behavior between social accounts in the social network, the account features corresponding to the target account are subjected to feature propagation mapping, so that the social network has an interactive behavior relationship with the target account. The account features of the social account are propagated and mapped to the target account, and the corresponding spread account features of the target account are obtained, and the spread account features are matched with the information interaction features of the information to be recommended, and the target information is determined according to the matching results for recommendation, thus using Information recommendation is performed based on the interactive behavior relationship between social accounts in the social network, which improves the accuracy of information recommendation.

在一个实施例中，如图3所示，特征传播映射的处理，即通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征，包括：In one embodiment, as shown in FIG. 3 , the processing of feature propagation mapping is to perform feature propagation mapping on the account features corresponding to the target account through account interaction features, and obtain the propagation account features corresponding to the target account according to the result of feature propagation mapping. ,include:

步骤302，从账号交互特征中提取与目标账号对应的目标账号交互特征。Step 302, extracting target account interaction features corresponding to the target account from the account interaction features.

其中，账号交互特征根据社交网络中各社交账号之间的交互行为得到，即账号交互特征包括社交网络中所有社交账号之间的交互行为。目标账号交互特征为社交网络中与目标账号存在交互行为的社交账号，与目标账号之间的交互特征，即目标账号交互特征为社交网络中与目标账号存在的交互行为所对应的特征。例如，如图4所示，在社交网络中，包括社交账号A、B、C、D和E，各社交账号可以作为社交网络中的节点，各节点之间通过边连接，节点连接的边表征社交账号之间的交互行为，边的宽度反映了社交账号之间交互的亲密程度，边越宽，表明社交账号之间的交互越亲密，即对应两个社交账号之间的交互行为越多或越频繁。在目标账号为社交账号A时，则从账号交互特征中提取得到的与社交账号A对应的目标账号交互特征包括特征A_AB、特征A_AD和特征A_AE，其中，特征A_AB为社交账号A与社交账号B之间的交互行为对应的交互特征，特征A_AD为社交账号A与社交账号D之间的交互行为对应的交互特征，特征A_AE为社交账号A与社交账号E之间的交互行为对应的交互特征。而社交账号A与社交账号C之间未产生交互行为，则不存在相应的交互特征。Wherein, the account interaction feature is obtained according to the interaction behavior between social accounts in the social network, that is, the account interaction feature includes the interaction behavior between all social accounts in the social network. The target account interaction feature is a social account in the social network that has interaction behavior with the target account, and the interaction feature with the target account, that is, the target account interaction feature is a feature corresponding to the interaction behavior with the target account in the social network. For example, as shown in Figure 4, in a social network, including social accounts A, B, C, D, and E, each social account can be used as a node in the social network, and each node is connected by an edge, and the edge of the node connection represents For the interaction between social accounts, the width of the edge reflects the intimacy of the interaction between the social accounts. The wider the edge, the more intimate the interaction between the social accounts, that is, the more or more interactions between the two social accounts. more often. When the target account is a social account A, the target account interaction features extracted from the account interaction features and corresponding to the social account A include feature A_AB, feature A_AD, and feature A_AE, wherein feature A_AB is social account A and social account B feature A_AD is the interaction feature corresponding to the interaction between social account A and social account D, and feature A_AE is the interaction feature corresponding to the interaction between social account A and social account E. However, if there is no interaction between the social account A and the social account C, there is no corresponding interaction feature.

具体地，在针对目标账号进行信息推荐时，服务器从社交网络对应的账号交互特征中提取目标账号对应的目标账号交互特征，目标账号交互特征包括根据目标账号与社交网络中其他社交账号之间的交互行为所得到的交互特征。Specifically, when recommending information for the target account, the server extracts the target account interaction features corresponding to the target account from the account interaction features corresponding to the social network. Interaction features obtained by the interaction behavior.

步骤304，获取目标账号交互特征对应的交互社交账号的账号特征。Step 304, acquiring the account characteristics of the interactive social account corresponding to the interaction characteristics of the target account.

其中，交互社交账号为社交网络中与目标账号具有交互行为的社交账号，即交互社交账号与目标账号交互特征对应，目标账号交互特征表征了目标账号与交互社交账号之间的交互行为的特征。例如，如图4所示，在目标账号为社交账号A时，账号交互特征中提取得到的与社交账号A对应的目标账号交互特征包括特征A_AB、特征A_AD和特征A_AE，则目标账号交互特征对应的交互社交账号包括社交账号B、D和E。交互社交账号的账号特征可以为表征交互社交账号特性的信息，具体可以为通过对交互社交账号进行特征工程处理得到。Wherein, the interactive social account is a social account that has interactive behavior with the target account in the social network, that is, the interactive social account corresponds to the interaction characteristics of the target account, and the interaction characteristics of the target account represent the characteristics of the interaction behavior between the target account and the interactive social account. For example, as shown in Figure 4, when the target account is a social account A, the target account interaction features extracted from the account interaction features corresponding to social account A include feature A_AB, feature A_AD and feature A_AE, then the target account interaction feature corresponds to The interactive social accounts of include social accounts B, D and E. The account feature of the interactive social account may be information characterizing the characteristics of the interactive social account, specifically, it may be obtained by performing feature engineering on the interactive social account.

在具体实现时，服务器可以预先对社交网络中的各社交账号进行特征工程处理，得到社交网络中的各社交账号各自的账号特征，并进行存储。在确定目标账号后，服务器可以直接从存储的账号特征中确定目标账号所对应的账号特征，进一步地，服务器可以从存储的账号特征中筛选出与目标账号存在交互行为的交互社交账号的账号特征。During specific implementation, the server may perform feature engineering processing on each social account in the social network in advance to obtain and store account features of each social account in the social network. After determining the target account, the server can directly determine the account features corresponding to the target account from the stored account features, and further, the server can filter out the account features of interactive social accounts that interact with the target account from the stored account features .

具体地，服务器确定目标账号交互特征后，根据目标账号交互特征获取与目标账号具有交互行为的交互社交账号对应的账号特征。Specifically, after the server determines the interaction characteristics of the target account, it acquires account characteristics corresponding to interactive social accounts that have interactive behavior with the target account according to the interaction characteristics of the target account.

步骤306，基于目标账号交互特征、交互社交账号的账号特征和传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果。Step 306 , based on the interaction features of the target account, the account features of the interactive social account, and the propagation mapping parameters, iterative feature propagation mapping is performed on the account features corresponding to the target account to obtain a result of feature propagation mapping.

其中，传播映射参数用于对特征传播映射进行调节，在通过网络模型实现特征传播映射时，传播映射参数可以通过网络模型训练确定。迭代是重复反馈过程的活动，而每一次迭代得到的结果会作为下一次迭代的初始值，即每次特征传播映射处理后，将得到的本次迭代特征传播映射的结果作为初始值进行下一次特征传播映射，直至迭代结束，得到特征传播映射的结果。在具体实现时，可以设置多个传播映射参数，通过目标账号交互特征、交互社交账号的账号特征和多个传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，如通过网络模型的多层结构分别进行迭代特征传播映射，得到特征传播映射的结果。Wherein, the propagation mapping parameter is used to adjust the feature propagation mapping, and when the feature propagation mapping is implemented through the network model, the propagation mapping parameter can be determined through network model training. Iteration is an activity that repeats the feedback process, and the result of each iteration will be used as the initial value of the next iteration, that is, after each feature propagation mapping process, the result of the iterative feature propagation mapping will be used as the initial value for the next iteration Feature propagation mapping, until the end of the iteration, get the result of feature propagation mapping. In actual implementation, multiple propagation mapping parameters can be set, and iterative feature propagation mapping is performed on the account characteristics corresponding to the target account through the interaction characteristics of the target account, the account characteristics of the interactive social account and multiple propagation mapping parameters, such as through the network model The multi-layer structure performs iterative feature propagation mapping respectively to obtain the result of feature propagation mapping.

具体地，服务器获取特征传播映射处理时的传播映射参数，基于目标账号交互特征、交互社交账号的账号特征和传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果。在一个具体应用中，特征传播映射通过预训练的特征传播映射模型实现，特征传播映射模型中包括至少一层层结构，每一层结构具有相应的传播映射参数，可以进行一次特征传播映射处理，通过特征传播映射模型可以对目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果。Specifically, the server obtains the propagation mapping parameters during feature propagation mapping processing, and performs iterative feature propagation mapping on the account characteristics corresponding to the target account based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters to obtain the feature propagation mapping. result. In a specific application, the feature propagation mapping is implemented through a pre-trained feature propagation mapping model, which includes at least one layer structure, each layer structure has a corresponding propagation mapping parameter, and a feature propagation mapping process can be performed once. Through the feature propagation mapping model, iterative feature propagation mapping can be performed on the account features corresponding to the target account to obtain the result of feature propagation mapping.

步骤308，根据特征传播映射的结果确定目标账号对应的传播账号特征。Step 308: Determine the characteristics of the propagation account corresponding to the target account according to the result of the characteristic propagation mapping.

得到特征传播映射的结果后，服务器可以根据特征传播映射的结果确定目标账号对应的传播账号特征，传播账号特征为目标账号对应的账号特征经过特征传播映射处理后获得的特征，传播账号特征除携带目标账号本身的特征外，还携带有社交网络中与目标账号存在社交关系的社交账号的账号特征。在具体实现时，服务器可以直接从特征传播映射的结果中提取得到目标账号对应的传播账号特征，也可以对特征传播映射的结果中包括的传播账号特征进行进一步特征处理，如进行归一化或标准化处理，得到目标账号对应的传播账号特征。After obtaining the result of feature propagation mapping, the server can determine the characteristics of the propagation account corresponding to the target account according to the result of characteristic propagation mapping. In addition to the characteristics of the target account itself, it also carries account characteristics of social accounts that have a social relationship with the target account in the social network. In specific implementation, the server can directly extract the propagation account features corresponding to the target account from the result of feature propagation mapping, or perform further feature processing on the propagation account features included in the result of feature propagation mapping, such as performing normalization or Standardize the process to obtain the characteristics of the broadcast account corresponding to the target account.

本实施例中，服务器基于目标账号交互特征、交互社交账号的账号特征和传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，并根据特征传播映射的结果确定目标账号对应的传播账号特征，从而通过交互行为对应的目标账号交互特征和传播映射参数，将社交网络中与目标账号具有交互行为的交互社交账号的账号特征传递至目标账号，获得能够准确表征社交网络中社交关系的传播账号特征，基于该传播账号特征进行信息推荐处理，可以提高信息推荐的准确性。In this embodiment, the server performs iterative feature propagation mapping on the account features corresponding to the target account based on the interaction features of the target account, the account features of the interactive social account, and the propagation mapping parameters, and determines the propagation account corresponding to the target account according to the result of the feature propagation mapping. feature, so that through the interaction characteristics of the target account corresponding to the interaction behavior and the propagation mapping parameters, the account characteristics of the interactive social account with the interaction behavior with the target account in the social network are transferred to the target account, and the communication that can accurately represent the social relationship in the social network is obtained. Account characteristics, information recommendation processing is performed based on the dissemination account characteristics, which can improve the accuracy of information recommendation.

在一个实施例中，基于目标账号交互特征、交互社交账号的账号特征和传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果，包括：将目标账号对应的账号特征确定为当前账号特征；基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果；将本次迭代特征传播映射的结果作为当前账号特征，并返回基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果的步骤。In one embodiment, based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters, iterative feature propagation mapping is performed on the account characteristics corresponding to the target account to obtain the result of the feature propagation mapping, including: The account feature is determined as the current account feature; based on the interaction feature of the target account, the account feature of the interactive social account and the propagation mapping parameters of this iterative feature propagation mapping, feature propagation mapping is performed on the current account feature to obtain the result of this iterative feature propagation mapping ; Use the result of this iterative feature propagation mapping as the current account feature, and return the propagation mapping parameters based on the target account interaction feature, the account feature of the interactive social account, and this iterative feature propagation mapping, and perform feature propagation mapping on the current account feature. The step of getting the result of this iterative feature propagation map.

其中，每一次迭代特征传播映射时，将上一次迭代特征传播映射的结果作为本次迭代特征传播映射的初始值，直至迭代完成，得到特征传播映射的结果。当前账号特征指本次进行特征传播映射时，目标账号对应的账号特征，在未进行特征传播映射时，当前账号特征取值为服务器获得的通过目标账号对应的账号特征。每次迭代特征传播映射可以设置有相应的传播映射参数，每次迭代特征传播映射对应的传播映射参数可以相同也可以不同，每次迭代特征传播映射对应的传播映射参数具体可以通过模型训练获得。Wherein, each time the feature propagation map is iterated, the result of the feature propagation map of the previous iteration is used as the initial value of the feature propagation map of this iteration, until the iteration is completed, and the result of the feature propagation map is obtained. The current account feature refers to the account feature corresponding to the target account when the feature propagation mapping is performed this time. When the feature propagation mapping is not performed, the current account feature value is the account feature corresponding to the target account obtained by the server. The corresponding propagation map parameters can be set for each iteration of the feature propagation map. The propagation map parameters corresponding to each iteration of the feature propagation map can be the same or different. The propagation map parameters corresponding to each iteration of the feature propagation map can be obtained through model training.

具体地，在对目标账号进行迭代特征传播映射时，服务器将目标账号对应的账号特征确定为当前账号特征，即将获得的目标账号对应的账号特征作为本次迭代特征传播映射的初始值。服务器基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，具体可以通过本次迭代特征传播映射的传播映射参数，将目标账号交互特征、交互社交账号的账号特征和当前账号特征进行融合映射，得到本次迭代特征传播映射的结果。服务器将本次迭代特征传播映射的结果作为下一次迭代特征传播映射的初始值，即将本次迭代特征传播映射的结果作为当前账号特征，返回基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果的步骤，从而实现对目标账号的账号特征进行迭代特征传播映射处理。Specifically, when performing iterative feature propagation mapping on the target account, the server determines the account feature corresponding to the target account as the current account feature, and uses the acquired account feature corresponding to the target account as the initial value of this iterative feature propagation mapping. Based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping, the server performs feature propagation mapping on the characteristics of the current account. Specifically, through the propagation mapping parameters of this iteration feature propagation mapping, the target The account interaction feature, the account feature of the interactive social account, and the current account feature are fused and mapped to obtain the result of this iterative feature propagation mapping. The server uses the result of this iterative feature propagation mapping as the initial value of the next iterative feature propagation mapping, that is, the result of this iterative feature propagation mapping as the current account feature, and returns the account features based on the target account interaction feature and interactive social account and this The propagation mapping parameter of the iterative feature propagation mapping, the step of performing feature propagation mapping on the characteristics of the current account, and obtaining the result of this iterative feature propagation mapping, so as to realize the iterative feature propagation mapping process on the account characteristics of the target account.

本实施例中，服务器通过每次迭代特征传播映射的传播映射参数，将目标账号交互特征、交互社交账号的账号特征和当前账号特征进行特征传播映射，并将每次迭代特征传播映射的结果作为下一次迭代特征传播映射的初始值，在迭代结束后得到特征传播映射的结果，从而通过迭代特征传播映射将社交网络中与目标账号具有交互行为的交互社交账号的账号特征从多维度传递至目标账号，获得能够准确表征社交网络中社交关系的传播账号特征，基于该传播账号特征进行信息推荐处理，可以提高信息推荐的准确性。In this embodiment, the server performs feature propagation mapping on the interaction features of the target account, the account features of the interactive social account, and the current account features through the propagation mapping parameters of each iteration of the feature propagation mapping, and uses the result of each iteration of the feature propagation mapping as The initial value of the feature propagation map for the next iteration, and the result of the feature propagation map is obtained after the iteration, so that the account features of the interactive social accounts in the social network that have interactive behavior with the target account in the social network are transferred from multiple dimensions to the target through the iterative feature propagation map The account number is used to obtain the characteristics of the communication account that can accurately represent the social relationship in the social network, and to perform information recommendation processing based on the characteristics of the communication account, which can improve the accuracy of the information recommendation.

在一个实施例中，基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果，包括：基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播，得到特征传播结果；对特征传播结果进行非线性映射，得到本次迭代特征传播映射的结果。In one embodiment, based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping, feature propagation mapping is performed on the current account characteristics to obtain the result of this iterative feature propagation mapping, including : Based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping, perform feature propagation on the current account characteristics to obtain the feature propagation results; perform nonlinear mapping on the feature propagation results to obtain this iteration The result of iterating over the feature propagation map.

其中，特征传播结果为通过目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播得到的结果。对特征传播结果进行非线性映射，可以增加特征传播映射处理中的非线性，从而可以提高特征表达的准确性，得到准确性高的传播账号特征。Wherein, the feature propagation result is the result obtained by performing feature propagation on the current account feature through the interaction feature of the target account, the account feature of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping. The non-linear mapping of the feature propagation results can increase the nonlinearity in the feature propagation mapping process, thereby improving the accuracy of feature expression and obtaining high-accuracy propagation account features.

具体地，服务器在基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播，得到特征传播结果后，如服务器通过本次迭代特征传播映射的传播映射参数，将目标账号交互特征、交互社交账号的账号特征和当前账号特征进行融合得到特征传播结果后，服务器对特征传播结果进行非线性映射，得到本次迭代特征传播映射的结果。具体可以由服务器获取预先设置的非线性函数，如Sigmoid函数、ReLU(Rectified Linear Unit，线性整流函数)函数等，对特征传播结果进行非线性映射处理，得到本次迭代特征传播映射的结果。在具体实现时，特征传播映射可以由特征传播映射模型实现，则在特征传播映射模型结构中，每一层特征传播映射的层结构，设置有相应的传播映射参数，以进行特征传播映射，在每次特征传播映射后，通过设置的激活层对每次特征传播映射的特征传播结果进行非线性映射，得到每次迭代特征传播映射的结果。激活层具体可以为ReLU层，从而得到本次迭代特征传播映射的结果。在得到本次迭代特征传播映射的结果后，进行下一次的特征传播映射，直至迭代结束，得到特征传播映射的结果，并根据特征传播映射的结果得到目标账号对应的传播账号特征。Specifically, based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping, the server performs feature propagation on the current account features, and after obtaining the feature propagation results, if the server passes this iteration feature The propagation mapping parameter of the propagation mapping, after combining the interaction characteristics of the target account, the account characteristics of the interactive social account and the current account characteristics to obtain the characteristic propagation result, the server performs nonlinear mapping on the characteristic propagation result to obtain the result of this iterative characteristic propagation mapping . Specifically, the server can obtain preset nonlinear functions, such as Sigmoid function, ReLU (Rectified Linear Unit, linear rectification function) function, etc., and perform nonlinear mapping processing on the feature propagation result to obtain the result of this iterative feature propagation mapping. In the specific implementation, the feature propagation mapping can be realized by the feature propagation mapping model, then in the feature propagation mapping model structure, the layer structure of each layer of feature propagation mapping is set with corresponding propagation mapping parameters to perform feature propagation mapping, in After each feature propagation mapping, the feature propagation result of each feature propagation mapping is nonlinearly mapped through the set activation layer to obtain the result of each iterative feature propagation mapping. The activation layer can specifically be a ReLU layer, so as to obtain the result of this iterative feature propagation mapping. After obtaining the result of this iterative feature propagation mapping, the next feature propagation mapping is performed until the end of the iteration, the result of the feature propagation mapping is obtained, and the propagation account characteristics corresponding to the target account are obtained according to the result of the feature propagation mapping.

本实施例中，在对目标账号对应的账号特征进行本次迭代特征传播映射，得到特征传播结果后，进一步对特征传播结果进行非线性映射，得到本次迭代特征传播映射的结果，从而对每次特征传播映射的结果进行非线性化处理，提高了特征传播映射的结果的非线性，可以增强获得的传播账号特征的特征表达，可以提高基于传播账号特征进行信息推荐的准确性。In this embodiment, after the iterative feature propagation mapping is performed on the account features corresponding to the target account to obtain the feature propagation result, the feature propagation result is further nonlinearly mapped to obtain the result of this iterative feature propagation mapping, so that each The result of the secondary feature propagation mapping is non-linearized, which improves the nonlinearity of the result of the feature propagation mapping, can enhance the feature expression of the obtained propagation account features, and can improve the accuracy of information recommendation based on the propagation account features.

在一个实施例中，如图5所示，特征传播映射的处理，即通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征，包括：In one embodiment, as shown in FIG. 5 , the processing of feature propagation mapping is to perform feature propagation mapping on the account features corresponding to the target account through account interaction features, and obtain the propagation account features corresponding to the target account according to the result of feature propagation mapping. ,include:

步骤502，确定根据社交网络中各社交账号的账号特征构建的网络账号特征；网络账号特征包括目标账号对应的账号特征。Step 502, determine the network account features constructed according to the account features of each social account in the social network; the network account features include the account features corresponding to the target account.

其中，网络账号特征根据社交网络中各社交账号的账号特征构建得到，网络账号特征包括社交网络中所有社交账号的账号特征，具体包括目标账号对应的账号特征，以及与目标账号存在交互行为的交互社交账号的账号特征。社交网络中各社交账号的账号特征可以基于特征工程实现，以对社交网络中各社交账号的特征进行准确的描述。Among them, the network account features are constructed according to the account features of each social account in the social network, and the network account features include the account features of all social accounts in the social network, specifically including the account features corresponding to the target account, and the interaction with the target account. Account characteristics of social accounts. The account characteristics of each social account in the social network can be implemented based on feature engineering, so as to accurately describe the characteristics of each social account in the social network.

具体地，在对目标账号对应的账号特征进行特征传播映射，确定目标账号对应的传播账号特征时，服务器可以直接对目标账号所属社交网络中的各社交账号统一进行特征传播映射处理，获得包括社交网络中各社交账号相应账号特征的网络账号特征。在具体应用时，服务器可以预先基于特征工程，提取得到目标账号所属社交网络中的各社交账号的账号特征，并构建得到社交网络对应的网络账号特征。例如，社交网络中的各社交账号可以作为社交网络的节点，而社交账号之间的交互行为可以作为连接节点的边，从而根据各社交账号之间的交互行为，构建社交网络对应的用户节点图，再通过节点嵌入算法，如Node2Vec算法基于用户节点图进行节点嵌入处理，得到社交网络对应的网络账号特征，网络账号特征中包括社交网络中各社交账号的账号特征。Specifically, when performing feature propagation mapping on the account features corresponding to the target account to determine the feature of the target account, the server can directly perform feature propagation mapping processing on all social accounts in the social network to which the target account belongs, and obtain information including social The network account characteristics corresponding to the account characteristics of each social account in the network. In a specific application, the server may extract account features of each social account in the social network to which the target account belongs based on feature engineering in advance, and construct network account features corresponding to the social network. For example, each social account in a social network can be used as a node of a social network, and the interaction behavior between social accounts can be used as an edge connecting nodes, so that a user node graph corresponding to a social network can be constructed according to the interaction behavior between social accounts , and then through the node embedding algorithm, such as the Node2Vec algorithm, based on the user node graph, the node embedding process is performed to obtain the network account characteristics corresponding to the social network, and the network account characteristics include the account characteristics of each social account in the social network.

步骤504，将账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射，获得特征传播映射模型输出的网络传播账号特征。Step 504, input the account interaction features and network account features into the feature propagation mapping model to perform feature propagation mapping, and obtain the network propagation account features output by the feature propagation mapping model.

其中，账号交互特征根据社交网络中各社交账号之间的交互行为得到，即账号交互特征包括社交网络中所有社交账号之间的交互行为。特征传播映射模型可以为机器学习模型，如可以为基于神经网络算法或深度学习算法训练得到的网络模型。特征传播映射模型可以预先训练得到，特征传播映射模型可以根据输入的账号交互特征和网络账号特征输入特征进行特征传播映射，并输出网络传播账号特征。网络传播账号特征包括社交网络中各社交账号分别对应的传播账号特征，网络传播账号特征由特征传播映射模型对社交网络的账号交互特征和网络账号特征进行特征传播映射，即对社交网络中的各社交账号统一进行特征传播映射，得到包括各社交账号相应传播账号特征的网络传播账号特征。Wherein, the account interaction feature is obtained according to the interaction behavior between social accounts in the social network, that is, the account interaction feature includes the interaction behavior between all social accounts in the social network. The feature propagation mapping model may be a machine learning model, such as a network model trained based on a neural network algorithm or a deep learning algorithm. The feature propagation mapping model can be obtained through pre-training, and the feature propagation mapping model can perform feature propagation mapping according to the input account interaction characteristics and network account feature input features, and output network propagation account features. The network communication account features include the communication account features corresponding to each social account in the social network. The network communication account features are mapped by the feature propagation mapping model to the social network account interaction features and network account features, that is, to each social network. The social accounts are uniformly mapped to feature propagation, and the characteristics of the network communication accounts including the characteristics of the corresponding communication accounts of each social account are obtained.

具体地，服务器可以查询预先通过神经网络算法或深度学习算法训练得到的特征传播映射模型，服务器将账号交互特征和网络账号特征输入特征传播映射模型中，以由特征传播映射模型进行特征传播映射，并获得特征传播映射模型输出的网络传播账号特征。Specifically, the server can query the feature propagation mapping model trained in advance through the neural network algorithm or deep learning algorithm, and the server inputs the account interaction features and network account features into the feature propagation mapping model to perform feature propagation mapping by the feature propagation mapping model, And the network communication account features output by the feature propagation mapping model are obtained.

步骤506，从网络传播账号特征中提取目标账号对应的传播账号特征。Step 506, extracting the characteristics of the communication account corresponding to the target account from the characteristics of the network communication account.

得到网络传播账号特征后，该网络传播账号特征中包括社交网络中所有社交账号的传播账号特征，则可以由服务器从网络传播账号特征中提取得到目标账号对应的传播账号特征。具体地，服务器可以根据目标账号与网络账号特征的映射关系，从特征传播映射模型输出的网络传播账号特征中，按照该映射关系提取得到目标账号对应的传播账号特征。After the network communication account features are obtained, the network communication account features include the communication account features of all social accounts in the social network, and the server can extract the communication account features corresponding to the target account from the network communication account features. Specifically, the server may extract, according to the mapping relationship between the target account and the network account features, the features of the target account corresponding to the target account from the feature propagation account features output by the feature propagation mapping model.

本实施例中，服务器通过预先训练完成的特征传播映射模型，基于社交网络的账号交互特征和网络账号特征进行特征传播映射，从而由特征传播映射模型对社交网络中的各社交账号进行特征传播映射，得到社交网络对应的网络传播账号特征，网络传播账号特征中包括社交网络中各社交账号对应的传播账号特征，服务器再从网络传播账号特征中提取得到目标账号对应的传播账号特征。通过特征传播映射模型进行特征传播映射，可以准确、快速地对社交网络中的各社交账号进行特征传播映射，有利于提高对社交网络中的社交账号进行信息推荐的处理效率和推荐准确性。In this embodiment, the server uses the pre-trained feature propagation mapping model to perform feature propagation mapping based on the account interaction characteristics of the social network and network account features, so that the feature propagation mapping model performs feature propagation mapping on each social account in the social network to obtain the network communication account features corresponding to the social network, the network communication account features include the communication account features corresponding to each social account in the social network, and the server then extracts the communication account features corresponding to the target account from the network communication account features. The feature propagation mapping through the feature propagation mapping model can accurately and quickly perform feature propagation mapping on each social account in the social network, which is conducive to improving the processing efficiency and recommendation accuracy of information recommendation for the social account in the social network.

在一个实施例中，将账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射，包括：通过标准化条件对账号交互特征进行标准化处理，得到标准化的账号交互特征；将标准化的账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射。In one embodiment, inputting account interaction features and network account features into a feature propagation mapping model to perform feature propagation mapping includes: performing standardized processing on account interaction features through standardized conditions to obtain standardized account interaction features; Features and network account features are input into the feature propagation mapping model for feature propagation mapping.

其中，标准化条件可以根据实际需要进行设置，具体可以包括特征标准化公式，通过对账号交互特征进行标准化处理，可以确保在通过特征传播映射模型进行特征传播映射时，能够控制输出的网络传播账号特征的维度数目，能够降低网络传播账号特征的复杂度，有利于提高信息推荐的处理效率。Among them, the standardization conditions can be set according to actual needs, which can specifically include feature standardization formulas. By standardizing the account interaction features, it can be ensured that when the feature propagation mapping is performed through the feature propagation mapping model, the output network propagation account characteristics can be controlled. The number of dimensions can reduce the complexity of account characteristics in network dissemination, which is conducive to improving the processing efficiency of information recommendation.

具体地，在得到账号交互特征和网络账号特征后，服务器获取预先设置的标准化条件，如账号交互特征包括账号交互特征矩阵时，可以通过账号交互特征矩阵的对角矩阵与账号交互特征矩阵的乘积，实现对账号交互特征矩阵的标准化处理，标准化条件可以根据实际需要进行灵活设置。服务器通过标准化条件对账号交互特征进行标准化处理，得到标准化的账号交互特征。服务器将标准化的账号交互特征和网络账号特征输入特征传播映射模型中，以由特征传播映射模型根据标准化的账号交互特征和网络账号特征进行特征传播映射，并输出网络传播账号特征。Specifically, after obtaining the account interaction features and network account features, the server obtains preset standardized conditions. For example, when the account interaction features include the account interaction feature matrix, the product of the diagonal matrix of the account interaction feature matrix and the account interaction feature matrix can be , to realize the standardized processing of the account interaction feature matrix, and the standardized conditions can be flexibly set according to actual needs. The server performs standardization processing on the account interaction features through standardized conditions to obtain standardized account interaction features. The server inputs the standardized account interaction features and network account features into the feature propagation mapping model, so that the feature propagation mapping model performs feature propagation mapping according to the standardized account interaction features and network account features, and outputs network propagation account features.

本实施例中，通过标准化条件对账号交互特征进行标准化处理，并由特征传播映射模型基于标准化的账号交互特征和网络账号特征进行特征传播映射，在特征传播映射模型进行多次特征传播映射处理时，可以有效控制数据的维度，降低特征传播映射模型输出的传播账号特征的复杂度，可以提高基于传播账号特征进行信息推荐的处理效率。In this embodiment, the account interaction features are standardized through the standardized conditions, and the feature propagation mapping is performed by the feature propagation mapping model based on the standardized account interaction characteristics and network account characteristics. When the feature propagation mapping model performs multiple feature propagation mapping processes , can effectively control the dimension of the data, reduce the complexity of the characteristics of the propagation account output by the feature propagation mapping model, and can improve the processing efficiency of information recommendation based on the characteristics of the propagation account.

在一个实施例中，如图6所示，推荐目标信息的处理，即将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐，包括：In one embodiment, as shown in FIG. 6 , the process of recommending target information is to match the characteristics of the dissemination account with each information interaction feature, and transfer the target information determined from each information to be recommended according to the matching result to the corresponding target account. The terminal makes recommendations, including:

步骤602，通过匹配模型对传播账号特征和各信息交互特征进行匹配，得到匹配模型输出的匹配结果。Instep 602, the characteristics of the dissemination account and the characteristics of each information interaction are matched by the matching model, and a matching result output by the matching model is obtained.

其中，匹配模型可以为机器学习模型，如可以为基于神经网络算法或深度学习算法训练得到的网络模型，如可以为多层感知机模型。匹配模型可以根据输入的传播账号特征和信息交互特征分别进行匹配，输出传播账号特征与信息交互特征的匹配结果，匹配结果可以反映传播账号特征与信息交互特征之间的相似程度，传播账号特征与信息交互特征的相似度越高，则传播账号特征与信息交互特征越匹配，则可以将信息交互特征对应的待推荐信息向该目标账号进行推荐。Wherein, the matching model may be a machine learning model, such as a network model trained based on a neural network algorithm or a deep learning algorithm, such as a multi-layer perceptron model. The matching model can perform matching according to the input characteristics of the communication account and information interaction characteristics, and output the matching results of the characteristics of the communication account and the information interaction characteristics. The matching result can reflect the similarity between the characteristics of the communication account and the information interaction characteristics. The higher the similarity of the information interaction features is, the more the features of the dissemination account match the information interaction features, and the information to be recommended corresponding to the information interaction features can be recommended to the target account.

具体地，服务器在将传播账号特征和各信息交互特征进行匹配时，服务器查询预先训练的匹配模型，服务器将传播账号特征和各信息交互特征依次输入到匹配模型中分别进行匹配，获得由匹配模型输出的匹配结果。Specifically, when the server matches the characteristics of the dissemination account and each information interaction characteristic, the server queries the pre-trained matching model, and the server sequentially inputs the characteristics of the dissemination account and each information interaction characteristic into the matching model for matching respectively, and obtains the information obtained by the matching model. Output matching results.

步骤604，基于匹配结果从各待推荐信息中确定目标信息。Instep 604, target information is determined from all pieces of information to be recommended based on the matching results.

其中，目标信息为根据匹配结果从各待推荐信息中确定的需要向目标账号进行推荐的信息，即目标信息为从各待推荐信息中筛选得到的向目标账号进行推荐的信息。具体地，得到匹配模型输出的匹配结果后，服务器根据匹配结果从各待推荐信息中确定目标信息。具体实现时，服务器可以将匹配结果与预设的推荐条件进行比较，从而确定满足推荐条件的匹配结果，并将满足推荐条件的匹配结果所对应的待推荐信息确定为目标信息。例如，匹配结果包括匹配相似度时，推荐条件可以为相似度阈值，则可以从匹配结果中确定匹配相似度大于相似度阈值的匹配结果，并将匹配相似度大于相似度阈值的匹配结果所对应的待推荐信息确定为目标信息。Wherein, the target information is the information that needs to be recommended to the target account determined from the information to be recommended according to the matching result, that is, the target information is the information to be recommended to the target account that is screened from the information to be recommended. Specifically, after obtaining the matching result output by the matching model, the server determines the target information from each information to be recommended according to the matching result. During specific implementation, the server may compare the matching result with the preset recommendation condition, thereby determining the matching result satisfying the recommendation condition, and determining the information to be recommended corresponding to the matching result satisfying the recommendation condition as the target information. For example, when the matching result includes matching similarity, the recommendation condition can be a similarity threshold, then it can be determined from the matching results that the matching similarity is greater than the similarity threshold, and the corresponding The information to be recommended is determined as the target information.

步骤606，将目标信息向目标账号对应的终端进行推荐。Step 606, recommending the target information to a terminal corresponding to the target account.

从各待推荐信息中确定目标信息后，服务器将目标信息向目标账号对应的终端进行推荐。具体可以由服务器根据目标信息生成信息推荐消息，并将信息推荐消息发送至目标账号对应的终端。在具体应用时，服务器可以在接收到目标账号对应的终端上传的信息推荐请求时，将目标信息向目标账号对应的终端进行推荐，也可以在满足信息推荐触发条件时，如达到信息推荐周期时，服务器主动向目标账号对应的终端推荐目标信息。After determining the target information from the information to be recommended, the server recommends the target information to the terminal corresponding to the target account. Specifically, the server may generate an information recommendation message according to the target information, and send the information recommendation message to a terminal corresponding to the target account. In a specific application, the server can recommend the target information to the terminal corresponding to the target account when receiving the information recommendation request uploaded by the terminal corresponding to the target account, or when the information recommendation trigger condition is met, such as when the information recommendation period is reached , the server actively recommends the target information to the terminal corresponding to the target account.

本实施例中，服务器通过预先训练的匹配模型对传播账号特征和各信息交互特征进行匹配，并根据匹配结果从各待推荐信息中确定目标信息向目标账号对应的终端进行推荐，可以实现传播账号特征和各信息交互特征的准确匹配，从而提高了信息推荐的准确性。In this embodiment, the server uses a pre-trained matching model to match the characteristics of the spreading account and the characteristics of each information interaction, and according to the matching result, determines the target information from the information to be recommended to recommend to the terminal corresponding to the target account, so that the spreading of the account can be realized. The accurate matching of features and each information interaction feature improves the accuracy of information recommendation.

在一个实施例中，基于匹配结果从各待推荐信息中确定目标信息包括：从匹配结果中确定满足推荐条件的目标匹配结果；将目标匹配结果所对应的待推荐信息确定为目标信息。In one embodiment, determining the target information from the information to be recommended based on the matching results includes: determining the target matching results satisfying the recommendation condition from the matching results; and determining the information to be recommended corresponding to the target matching results as the target information.

其中，推荐条件可以根据实际需要进行灵活设置，如可以设置为匹配程度大于匹配程度阈值，或满足匹配程度排序要求等。具体地，服务器获得匹配模型输出的匹配结果后，服务器进一步获取预设的推荐条件，根据推荐条件对匹配结果进行筛选，从而从匹配结果中确定满足推荐条件的目标匹配结果，并确定目标匹配结果对应的待推荐信息，服务器将目标匹配结果所对应的待推荐信息确定为目标信息，以将目标信息向目标账号对应的终端进行推荐。Wherein, the recommendation condition can be flexibly set according to actual needs, for example, it can be set that the matching degree is greater than a matching degree threshold, or meets the matching degree sorting requirement. Specifically, after the server obtains the matching result output by the matching model, the server further obtains the preset recommendation condition, and filters the matching result according to the recommendation condition, thereby determining the target matching result that satisfies the recommendation condition from the matching result, and determines the target matching result For the corresponding information to be recommended, the server determines the information to be recommended corresponding to the target matching result as the target information, so as to recommend the target information to the terminal corresponding to the target account.

在一个具体应用中，推荐条件为匹配程度大于匹配程度阈值，则服务器可以将匹配结果与预设的匹配程度阈值进行比较，将匹配程度大于匹配程度阈值的匹配结果确定为目标匹配结果，并将目标匹配结果所对应的待推荐信息确定为目标信息。又一个具体应用中，推荐条件为匹配程度最高的20个匹配结果，则服务器可以将匹配结果按照匹配程度从高到低进行排序，服务器取排序前20的匹配结果作为目标匹配结果，并将目标匹配结果所对应的待推荐信息确定为目标信息。In a specific application, the recommendation condition is that the matching degree is greater than the matching degree threshold, then the server can compare the matching result with the preset matching degree threshold, determine the matching result with the matching degree greater than the matching degree threshold as the target matching result, and The information to be recommended corresponding to the target matching result is determined as the target information. In another specific application, if the recommendation condition is the 20 matching results with the highest matching degree, the server can sort the matching results according to the matching degree from high to low, and the server takes the top 20 matching results as the target matching results, and sets the target The information to be recommended corresponding to the matching result is determined as the target information.

本实施例中，通过预设的推荐条件对匹配结果进行筛选，将筛选得到的满足推荐条件的匹配结果确定为目标匹配结果，从而根据目标匹配结果从待推荐信息中确定目标信息，可以基于推荐条件对匹配结果进行筛选，从而选择与目标账号匹配的目标信息进行推荐，可以提高信息推荐的准确性。In this embodiment, the matching results are screened through the preset recommendation conditions, and the filtered matching results meeting the recommendation conditions are determined as the target matching results, so that the target information is determined from the information to be recommended according to the target matching results. The conditions filter the matching results, so that the target information matching the target account is selected for recommendation, which can improve the accuracy of information recommendation.

在一个实施例中，信息推荐方法还包括：确定目标账号所属的社交网络中各社交账号之间交互行为的统计结果；基于统计结果获得社交网络中各社交账号之间的交互特征；根据社交网络中各社交账号之间的交互特征，生成社交网络对应的账号交互特征。In one embodiment, the information recommendation method further includes: determining statistical results of interaction behaviors between social accounts in the social network to which the target account belongs; obtaining interaction features between social accounts in the social network based on the statistical results; The interaction features between social accounts in the social network are generated to generate account interaction features corresponding to the social network.

其中，目标账号为社交网络中需要进行信息推荐的用户账号，账号交互特征根据社交网络中各社交账号之间的交互行为得到，交互行为包括用户通过各社交账号进行的聊天、评论、点赞、转发等各种交互操作。统计结果为对各社交账号之间的交互行为进行统计获得的统计数据，如可以包括社交账号之间交互行为的累积次数、交互频率、交互累积时长等。统计结果可以根据实际需要进行设置。交互特征用于表征社交网络中两两社交账号之间的交互行为，具体可以基于社交网络中两两社交账号之间交互行为的统计结果生成，如可以对两两社交账号之间交互行为的统计结果进行特征提取，得到两两社交账号之间的交互特征。Among them, the target account is a user account that needs information recommendation in the social network, and the account interaction feature is obtained according to the interaction behavior between the social accounts in the social network, and the interaction behavior includes chatting, commenting, liking, Various interactive operations such as forwarding. The statistics result is statistical data obtained by counting the interaction behaviors between social accounts, for example, it may include the cumulative number of interactions between social accounts, the interaction frequency, the cumulative interaction duration, and the like. Statistical results can be set according to actual needs. Interaction features are used to characterize the interaction behavior between two social accounts in a social network, specifically, it can be generated based on the statistical results of the interaction behavior between two social accounts in a social network, for example, the statistics of the interaction behavior between two social accounts can be As a result, feature extraction is performed to obtain the interaction features between two social accounts.

具体地，服务器获取目标账号所属的社交网络中各社交账号之间交互行为的统计结果，统计结果可以根据对各社交账号之间交互行为进行统计处理得到，如统计各社交账号之间交互行为的累积次数。服务器基于各社交账号之间交互行为的统计结果，确定各社交账号之间的交互特征，具体可以由服务器对各社交账号之间交互行为的统计结果进行特征化处理，得到各社交账号之间的交互特征。得到社交网络中各社交账号之间的交互特征后，服务器根据各社交账号之间的交互特征生成社交网络对应的账号交互特征，账号交互特征包括各社交账号之间的交互特征。在具体实现时，服务器可以将各社交账号之间的交互特征进行组合，生成社交网络对应的账号交互特征。Specifically, the server obtains statistical results of the interaction behaviors between social accounts in the social network to which the target account belongs, and the statistical results can be obtained by statistically processing the interaction behaviors between the social accounts, for example, the statistics of the interaction behaviors between the social accounts Cumulative times. The server determines the interaction characteristics between the social accounts based on the statistical results of the interaction behaviors between the social accounts. Specifically, the server can characterize the statistical results of the interaction behaviors between the social accounts to obtain the interaction characteristics between the social accounts. interactive features. After obtaining the interaction features between the social accounts in the social network, the server generates account interaction features corresponding to the social network according to the interaction features between the social accounts, where the account interaction features include the interaction features between the social accounts. During specific implementation, the server may combine the interaction features between social accounts to generate account interaction features corresponding to the social network.

本实施例中，通过社交网络中各社交账号之间交互行为的统计结果，确定各社交账号之间的交互特征，并基于各社交账号之间的交互特征生成社交网络对应的账号交互特征，基于社交网络中各社交账号间的交互行为构建社交网络对应的账号交互特征，从而可以通过账号交互特征对各社交账号的账号特征进行特征传播映射，基于特征传播映射获得的传播账号特征可以实现对目标账号的准确信息推荐。In this embodiment, through the statistical results of the interaction behavior between the social accounts in the social network, the interaction features between the social accounts are determined, and the account interaction features corresponding to the social network are generated based on the interaction features between the social accounts, based on The interaction behavior between social accounts in the social network constructs the corresponding account interaction characteristics of the social network, so that the account characteristics of each social account can be mapped through the account interaction characteristics, and the propagation account characteristics obtained based on the characteristic propagation mapping can realize the target The accurate information of the account is recommended.

在一个实施例中，信息推荐方法还包括：确定目标账号所属的社交网络；对社交网络中的各节点进行节点嵌入，得到各节点分别对应的节点特征；各节点与社交网络中的各社交账号对应，各节点之间的节点关系与各社交账号之间的交互行为对应；基于各节点对应的节点特征，确定目标账号对应的账号特征。In one embodiment, the information recommendation method further includes: determining the social network to which the target account belongs; performing node embedding on each node in the social network to obtain node characteristics corresponding to each node; Corresponding, the node relationship between each node corresponds to the interaction behavior between each social account; based on the node characteristics corresponding to each node, the account characteristics corresponding to the target account are determined.

其中，目标账号为需要进行信息推荐的用户账号，社交网络为目标账号所属社交体系对应的网络，社交网络由各用户对应的社交账号构成。社交网络中的社交账号可以作为社交网络的节点，每一社交账号对应于一个用户，各节点之间的交互行为可以通过连接节点的边进行描述，从而根据各社交账号之间的交互行为，构建社交网络对应的用户节点图。基于用户节点图可以通过嵌入算法，如Node2Vec算法进行节点嵌入处理，得到用户节点图中各节点分别对应的节点特征。根据节点对应的节点特征，可以得到目标账号对应的账号特征，如可以将目标账号对应节点的节点特征作为目标账号对应的账号特征。Wherein, the target account is a user account that needs information recommendation, the social network is a network corresponding to the social system to which the target account belongs, and the social network is composed of social accounts corresponding to each user. The social accounts in the social network can be used as the nodes of the social network, each social account corresponds to a user, and the interaction between nodes can be described by the edges connecting the nodes, so that according to the interaction between the social accounts, construct The user node graph corresponding to the social network. Based on the user node graph, an embedding algorithm, such as the Node2Vec algorithm, can be used to perform node embedding processing to obtain node features corresponding to each node in the user node graph. According to the node feature corresponding to the node, the account feature corresponding to the target account can be obtained, for example, the node feature of the node corresponding to the target account can be used as the account feature corresponding to the target account.

具体地，服务器确定目标账号所属的社交网络，将社交网络中的社交账号映射为相应的节点，并将社交账号之间的交互行为映射为节点之间的边，在社交账号之间存在交互行为时，通过边连接节点，边的宽度与交互行为的统计结果对应。如图7所示，在社交网络中，包括社交账号A、B、C、D和E，图7中记录了在一定时间段内，社交账号之间的交互行为的统计数据，包括聊天次数、交互累积时长、点赞和评论操作的次数等。如图8所示，将图7中的各社交账号A、B、C、D和E映射为节点A、B、C、D和E，各节点通过边连接，边的宽度反映了社交账号之间的交互行为的统计数据，统计数据具体可以实际需要设置各种类型交互行为的权值，通过加权处理得到。节点A、B、C、D和E与社交账号A、B、C、D和E一一对应，各节点之间的节点关系，如节点的连接关系以及连接的边的宽度，与社交账号之间的交互行为对应。Specifically, the server determines the social network to which the target account belongs, maps the social accounts in the social network to corresponding nodes, and maps the interaction between social accounts to edges between nodes, and there is an interaction between social accounts When , the nodes are connected by edges, and the width of the edges corresponds to the statistical results of the interaction behavior. As shown in Figure 7, the social network includes social accounts A, B, C, D, and E. Figure 7 records the statistical data of the interactive behavior between social accounts within a certain period of time, including the number of chats, The cumulative duration of interactions, the number of likes and comments, etc. As shown in Figure 8, the social accounts A, B, C, D, and E in Figure 7 are mapped to nodes A, B, C, D, and E, and each node is connected by an edge, and the width of the edge reflects the relationship between the social accounts. The statistical data of the interactive behavior among them can be obtained through weighting processing by setting the weights of various types of interactive behaviors according to actual needs. Nodes A, B, C, D, and E are in one-to-one correspondence with social accounts A, B, C, D, and E. The node relationship between nodes, such as the connection relationship between nodes and the width of connected edges, is related to the relationship between social accounts. Correspondence between interactive behaviors.

服务器针对社交网络中的各节点进行节点嵌入，如通过Node2Vec节点嵌入算法对各节点进行节点嵌入，得到各节点分别对应的节点特征。其中，嵌入可以将实体映射到连续的向量空间中，使得实体可以用向量来表示，而节点嵌入可以通过某个节点的嵌入特征找到它在节点图中的邻居节点，并可以将该嵌入特征描述对应的节点。服务器基于各节点对应的节点特征，确定目标账号对应的账号特征。具体实现时，可以由服务器根据目标账号与节点的映射关系，从社交网络中的各节点中确定目标节点，并根据节点嵌入的结果获取目标节点对应的目标节点特征，服务器将目标节点特征确定为目标账号对应的账号特征。The server performs node embedding for each node in the social network, for example, performs node embedding for each node through the Node2Vec node embedding algorithm, and obtains node features corresponding to each node. Among them, embedding can map entities into a continuous vector space, so that entities can be represented by vectors, and node embedding can find its neighbor nodes in the node graph through the embedding features of a node, and can describe the embedding features the corresponding node. Based on the node characteristics corresponding to each node, the server determines the account characteristics corresponding to the target account. During specific implementation, the server can determine the target node from each node in the social network according to the mapping relationship between the target account and the node, and obtain the target node characteristics corresponding to the target node according to the node embedding result, and the server determines the target node characteristics as The account characteristics corresponding to the target account.

本实施例中，通过对与社交网络中的社交账号对应的节点进行节点嵌入处理，可以构建能够准确表达社交账号的账号特征，基于该账号特征进行信息推荐，可以提高信息推荐的准确性。In this embodiment, by performing node embedding processing on nodes corresponding to social accounts in the social network, account features that can accurately express social accounts can be constructed, and information recommendation based on the account features can improve the accuracy of information recommendation.

在一个实施例中，对社交网络中的各节点进行节点嵌入，得到各节点分别对应的节点特征，包括：根据社交网络中各节点之间的节点关系，确定各节点之间的游走权重；以每个节点为起点，基于游走权重在社交网络中进行节点游走，形成各节点游走轨迹；通过嵌入模型对各节点游走轨迹进行特征嵌入，得到各节点分别对应的节点特征。In one embodiment, the node embedding is performed on each node in the social network to obtain the node characteristics corresponding to each node, including: determining the walking weight between each node according to the node relationship between each node in the social network; Taking each node as the starting point, the node walks in the social network based on the walking weight to form the walking trajectory of each node; the feature embedding of each node's walking trajectory is carried out through the embedding model, and the node characteristics corresponding to each node are obtained.

其中，游走权重与节点之间的节点关系对应，节点关系反映了社交账号之间的交互行为，社交账号之间的交互行为越丰富、越频繁，则相应的节点关系越亲密，游走权重数值越大。节点游走轨迹为以社交网络中的节点为起点，根据连接节点的边进行游走得到的轨迹。嵌入模型可以对节点游走轨迹进行特征映射，具体可以为基于word2vec词向量嵌入算法预先构建的网络模型，嵌入模型可以对输入的节点游走轨迹进行特征嵌入处理，将节点游走轨迹映射成相应节点的节点特征，节点特征反映了相应节点在社交网络中的交互关系。Among them, the walking weight corresponds to the node relationship between nodes, and the node relationship reflects the interaction between social accounts. The richer and more frequent the interaction between social accounts, the closer the corresponding node relationship is, and the walking weight The larger the value is. The node walking trajectory is a trajectory obtained by taking a node in the social network as a starting point and walking according to the edges connecting the nodes. The embedding model can perform feature mapping on node walking trajectories. Specifically, it can be a network model pre-built based on the word2vec word vector embedding algorithm. The embedding model can perform feature embedding processing on the input node walking trajectories, and map node walking trajectories into corresponding The node characteristics of the node, the node characteristics reflect the interaction relationship of the corresponding node in the social network.

具体地，服务器确定社交网络中各节点之间的节点关系，节点关系与各社交账号之间的交互行为对应，服务器根据节点关系确定相应节点之间的游走权重。例如，可以预先建立节点关系与游走权重之间的映射关系，从而可以根据该映射关系查询得到各节点的节点关系所对应的游走权重。服务器以每个节点为起点，基于游走权重在社交网络的各节点间进行节点游走，具体根据节点间连接的边在节点间进行随机游走，形成各节点游走轨迹。服务器通过预先训练的嵌入模型，对获得的各节点游走轨迹分别进行特征嵌入处理，具体可以由服务器将各节点游走轨迹分别输入嵌入模型中，由嵌入模型进行特征嵌入处理，并输出各节点分别对应的节点特征。Specifically, the server determines the node relationship between the nodes in the social network, the node relationship corresponds to the interaction behavior between the social accounts, and the server determines the roaming weight between corresponding nodes according to the node relationship. For example, a mapping relationship between a node relationship and a walking weight may be established in advance, so that the walking weight corresponding to the node relationship of each node may be queried according to the mapping relationship. Starting from each node, the server performs node walk among the nodes of the social network based on the walk weight, and specifically performs random walk among the nodes according to the edges connected between the nodes to form the walk trajectory of each node. The server uses the pre-trained embedding model to perform feature embedding processing on the obtained walking trajectories of each node. Specifically, the server can input the walking trajectories of each node into the embedding model, and the embedding model performs feature embedding processing and outputs each node. Corresponding node features respectively.

本实施例中，通过节点关系确定的游走权重，使各节点在社交网络中进行带权游走，并通过嵌入模型对获得的各节点游走轨迹进行特征嵌入，可以获得基于节点嵌入算法的节点特征，节点特征反映了社交网络中社交账号之间的交互行为，可以对社交账号在社交网络的社交关系进行准确表达，基于社交账号的账号特征进行信息推荐，可以提高信息推荐的准确性。In this embodiment, through the walking weight determined by the node relationship, each node is allowed to carry out weighted walking in the social network, and the feature embedding of each node's walking trajectory obtained through the embedding model can be obtained based on the node embedding algorithm. Node features, which reflect the interaction between social accounts in social networks, can accurately express the social relationships of social accounts in social networks, and recommend information based on account features of social accounts, which can improve the accuracy of information recommendations.

在一个实施例中，信息推荐方法还包括：确定每个待推荐信息对各自的交互社交账号；交互社交账号为社交网络中对相应待推荐信息产生过交互操作的社交账号；对各交互社交账号对应的账号特征进行特征聚合，得到相应待推荐信息的信息交互特征。In one embodiment, the information recommendation method further includes: determining each interactive social account for each information to be recommended pair; the interactive social account is a social account that has interacted with the corresponding information to be recommended in the social network; The corresponding account features are aggregated to obtain the information interaction features of the corresponding information to be recommended.

其中，待推荐信息为能够进行推荐的信息，具体可以为各种类型的信息，如音视频、图片、文本、网页、名片等各种数据信息。信息交互特征为每个待推荐信息的特征，信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的。Wherein, the information to be recommended is information that can be recommended, and specifically may be various types of information, such as various data information such as audio and video, pictures, texts, web pages, and business cards. The information interaction feature is a feature of each information to be recommended, and the information interaction feature is generated based on account features of social accounts that have interacted with the corresponding information to be recommended in the social network.

具体地，服务器确定每个待推荐信息对各自的交互社交账号，交互社交账号为社交网络中对相应待推荐信息产生过交互操作的社交账号。例如，社交网络中，社交账号A、B和D均对待推荐信息R产生过交互操作，如浏览过待推荐信息R，或对待推荐信息R进行点赞操作过，则对于待推荐信息R，其交互社交账号包括社交账号A、B和D，而对于社交账号C，其未对待推荐信息R产生过交互操作，待推荐信息R的交互社交账号不包括社交账号C。Specifically, the server determines an interactive social account for each pair of information to be recommended, and the interactive social account is a social account that has interacted with the corresponding information to be recommended in a social network. For example, in a social network, social accounts A, B, and D have all interacted with the information to be recommended R. The interactive social accounts include social accounts A, B, and D, and for the social account C, it has not interacted with the recommended information R, and the interactive social accounts of the recommended information R do not include the social account C.

服务器确定各交互社交账号对应的账号特征，并将各交互社交账号对应的账号特征进行特征聚合，得到相应待推荐信息的信息交互特征。在具体实现时，服务器可以对各交互社交账号对应的账号特征进行融合，如可以进行平均池化处理，得到待推荐信息的信息交互特征，信息交互特征综合了对待推荐信息产生过交互操作的交互社交账号的账号特征，从而通过交互社交账号对应的账号特征对待推荐信息的特征进行描述，以便基于社交网络中的社交关系进行信息推荐处理。The server determines account features corresponding to each interactive social account, and aggregates the account features corresponding to each interactive social account to obtain information interaction features corresponding to the information to be recommended. In actual implementation, the server can fuse the account features corresponding to each interactive social account, for example, average pooling can be performed to obtain the information interaction features of the information to be recommended. The account features of the social account, so as to describe the features of the information to be recommended by interacting with the account features corresponding to the social account, so as to perform information recommendation processing based on the social relationship in the social network.

本实施例中，根据对待推荐信息产生过交互操作的交互社交账号的账号特征进行特征聚合，获得相应待推荐信息的信息交互特征，从而可以融合交互社交账号对应的账号特征，通过交互社交账号对应的账号特征对待推荐信息的特征进行描述，以便基于社交网络中的社交关系进行信息推荐处理，实现基于社网络中的社交关系进行信息推荐，提高了信息推荐的准确性。In this embodiment, the feature aggregation is performed according to the account features of the interactive social accounts that have generated interactive operations on the information to be recommended, and the information interaction features of the corresponding information to be recommended are obtained, so that the account features corresponding to the interactive social accounts can be fused. The account characteristics of the account describe the characteristics of the information to be recommended, so as to perform information recommendation processing based on the social relationship in the social network, realize information recommendation based on the social relationship in the social network, and improve the accuracy of information recommendation.

在一个实施例中，如图9所示，提供了一种信息推荐方法，以该方法应用于图1中的服务器为例进行说明，包括以下步骤：In one embodiment, as shown in FIG. 9, a method for recommending information is provided. The method is applied to the server in FIG. 1 as an example for illustration, including the following steps:

步骤902，获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到。Step 902, acquiring the account interaction features corresponding to the social network to which the target account belongs; the account interaction features are obtained according to the interaction behavior between social accounts in the social network.

本实施例中，账号交互特征根据社交网络中各社交账号之间的交互行为得到，具体包括：确定目标账号所属的社交网络中各社交账号之间交互行为的统计结果；基于统计结果获得社交网络中各社交账号之间的交互特征；根据社交网络中各社交账号之间的交互特征，生成社交网络对应的账号交互特征。账号交互特征反映了社交网络中各社交账号支架的交互关系。In this embodiment, the account interaction features are obtained according to the interaction behaviors between social accounts in the social network, specifically including: determining the statistical results of the interaction behaviors between social accounts in the social network to which the target account belongs; obtaining social network information based on the statistical results. Interaction features between social accounts in the social network; according to the interaction features between social accounts in the social network, account interaction features corresponding to the social network are generated. The account interaction features reflect the interaction relationship of each social account bracket in the social network.

步骤904，确定根据社交网络中各社交账号的账号特征构建的网络账号特征；网络账号特征包括目标账号对应的账号特征；Step 904, determining the network account characteristics constructed according to the account characteristics of each social account in the social network; the network account characteristics include the account characteristics corresponding to the target account;

步骤906，通过标准化条件对账号交互特征进行标准化处理，得到标准化的账号交互特征；Step 906, standardize the account interaction features through standardized conditions to obtain standardized account interaction features;

步骤908，将标准化的账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射，获得特征传播映射模型输出的网络传播账号特征；Step 908, input the standardized account interaction features and network account features into the feature propagation mapping model to perform feature propagation mapping, and obtain the network propagation account characteristics output by the feature propagation mapping model;

步骤910，从网络传播账号特征中提取目标账号对应的传播账号特征。Step 910, extracting the characteristics of the communication account corresponding to the target account from the characteristics of the network communication account.

本实施例中，网络账号特征根据社交网络中各社交账号的账号特征构建得到，网络账号特征包括社交网络中所有社交账号的账号特征，具体包括目标账号对应的账号特征，以及与目标账号存在交互行为的交互社交账号的账号特征。账号交互特征包括账号交互特征矩阵，标准化条件为通过账号交互特征矩阵的对角矩阵与账号交互特征矩阵的乘积的标准化处理规则。特征传播映射模型为预先基于神经网络算法训练的机器学习模型，特征传播映射模型可以根据输入的账号交互特征和网络账号特征输入特征进行特征传播映射，并输出包括各社交账号相应传播账号特征的网络传播账号特征，传播账号特征除携带目标账号本身的特征外，还携带有社交网络中与目标账号存在社交关系的社交账号的账号特征。In this embodiment, the network account features are constructed according to the account features of each social account in the social network. The network account features include account features of all social accounts in the social network, specifically account features corresponding to the target account, and interactions with the target account. The account characteristics of the interactive social account of the behavior. The account interaction feature includes an account interaction feature matrix, and the normalization condition is a normalization processing rule based on a product of a diagonal matrix of the account interaction feature matrix and an account interaction feature matrix. The feature propagation mapping model is a machine learning model pre-trained based on the neural network algorithm. The feature propagation mapping model can perform feature propagation mapping according to the input account interaction characteristics and network account characteristics input features, and output a network including the corresponding propagation account characteristics of each social account. The characteristics of the dissemination account, the characteristics of the dissemination account carry not only the characteristics of the target account itself, but also the account characteristics of social accounts that have a social relationship with the target account in the social network.

步骤912，获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的。Step 912, acquiring the respective information interaction characteristics of each information to be recommended; the information interaction characteristics are generated based on the account characteristics of social accounts in the social network that have interacted with the corresponding information to be recommended.

本实施例中，待推荐信息为需要进行推荐的文本、网页、图片或音视频等数据，信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的，用于描述每个待推荐信息的特征。In this embodiment, the information to be recommended is data such as text, web pages, pictures, or audio and video that need to be recommended, and the information interaction features are generated based on the account features of social accounts that have interacted with the corresponding information to be recommended in the social network. It is used to describe the characteristics of each information to be recommended.

步骤914，通过匹配模型对传播账号特征和各信息交互特征进行匹配，得到匹配模型输出的匹配结果；Step 914, matching the characteristics of the dissemination account and the characteristics of each information interaction through the matching model, and obtaining the matching result output by the matching model;

步骤916，从匹配结果中确定满足推荐条件的目标匹配结果；Step 916, determine the target matching result satisfying the recommendation condition from the matching result;

步骤918，将目标匹配结果所对应的待推荐信息确定为目标信息；Step 918, determining the information to be recommended corresponding to the target matching result as the target information;

步骤920，将目标信息向目标账号对应的终端进行推荐。Step 920, recommending the target information to a terminal corresponding to the target account.

本实施例中，匹配模型为基于神经网络算法预先训练的多层感知机模型，匹配模型可以根据输入的传播账号特征和信息交互特征分别进行匹配，输出传播账号特征与信息交互特征的匹配结果，匹配结果可以反映传播账号特征与信息交互特征之间的相似程度，传播账号特征与信息交互特征的相似度越高，则传播账号特征与信息交互特征越匹配，则可以将信息交互特征对应的待推荐信息向该目标账号进行推荐。推荐条件为匹配程度大于匹配程度阈值，目标匹配结果为匹配程度大于匹配程度阈值的匹配结果，目标信息为目标匹配结果对应的待推荐信息，确定目标信息后，服务器将目标信息向目标账号对应的终端进行推荐。In this embodiment, the matching model is a multi-layer perceptron model pre-trained based on a neural network algorithm. The matching model can perform matching according to the input communication account features and information interaction features, and output the matching results of the communication account features and information interaction features. The matching result can reflect the degree of similarity between the characteristics of the dissemination account and the characteristics of information interaction. The higher the similarity between the characteristics of the dissemination account and the characteristics of information interaction, the more matching the characteristics of the dissemination account and the characteristics of information interaction. The recommendation information is used to recommend to the target account. The recommendation condition is that the matching degree is greater than the matching degree threshold, the target matching result is the matching result whose matching degree is greater than the matching degree threshold, and the target information is the information to be recommended corresponding to the target matching result. After the target information is determined, the server sends the target information to the corresponding target account. The terminal makes recommendations.

本实施例中，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，从而将社交网络中与目标账号具有交互行为关系的社交账号的账号特征传播映射至目标账号，得到目标账号对应的传播账号特征，并通过传播账号特征和待推荐信息各自的信息交互特征进行匹配，根据匹配结果确定目标信息进行推荐，从而利用了社交网络中各社交账号之间的交互行为关系进行信息推荐，提高了信息推荐的准确性。In this embodiment, through the account interaction features obtained according to the interaction behavior between social accounts in the social network, feature propagation mapping is performed on the account features corresponding to the target account, so that the social network that has an interactive behavior relationship with the target account in the social network The account feature propagation of the account is mapped to the target account, and the corresponding propagation account characteristics of the target account are obtained, and the characteristics of the propagating account are matched with the information interaction characteristics of the information to be recommended, and the target information is determined according to the matching result for recommendation, thus making use of the social network The interactive behavior relationship between social accounts in the network is used to recommend information, which improves the accuracy of information recommendation.

本申请还提供一种应用场景，该应用场景应用上述的信息推荐方法。具体地，该信息推荐方法在该应用场景的应用如下：The present application also provides an application scenario, where the above-mentioned information recommendation method is applied. Specifically, the application of the information recommendation method in this application scenario is as follows:

该应用场景中需要进行推荐的信息为短视频，短视频即短片视频，是一种互联网内容传播方式，一般是在互联网新媒体上传播的时长在30分钟以内的视频；随着移动终端普及和网络的提速，短平快的大流量传播内容逐渐获得各大平台、粉丝和资本的青睐。The information that needs to be recommended in this application scenario is a short video, which is a short video, which is a way of Internet content dissemination, generally a video that is less than 30 minutes in length on new Internet media; with the popularization of mobile terminals and With the speeding up of the Internet, short, flat, fast and high-traffic communication content has gradually won the favor of major platforms, fans and capital.

由于短视频娱乐的兴起时间短，短视频推荐成为各大互联网公司寸土必争的主战场。尽管传统处理中通过协同过滤等算法进行短视频推荐，但是对于某些后入场的平台，由于其积累较少，尽管具有大量的用户社交网络信息，用户实际对短视频的浏览行为较少，导致直接使用成熟算法进行短视频推荐的效果不佳。例如，传统的短视频推荐处理中，一种方式为基于用户相似度的推荐方法，对于一个给定用户，通过推荐与该用户相似的用户浏览过的短视频给出推荐结果；另一种方式为基于短视频的推荐方法，给定一个用户，通过推荐与该用户浏览视频最相似的视频的方法给出推荐结果。此外，传统推荐处理中，还可以基于匹配度的推荐技术主要通过计算给定用户与候选集视频的匹配度，然后按照匹配度对候选集视频进行排序，随后推荐匹配度最高的几个视频。然而，传统的短视频推荐处理中，能够得到准确推荐的前提需要积累足够的用户浏览点击数据，而在用户的浏览数据量较少时，无法进行准确的短视频推荐。Due to the short-term rise of short video entertainment, short video recommendation has become the main battlefield for major Internet companies to compete for every inch of land. Although in the traditional processing, algorithms such as collaborative filtering are used to recommend short videos, for some platforms that enter the market later, due to their less accumulation and a large amount of user social network information, users actually browse less for short videos. As a result, the effect of directly using mature algorithms for short video recommendation is not good. For example, in the traditional short video recommendation process, one method is the recommendation method based on user similarity. For a given user, the recommendation results are given by recommending short videos viewed by users similar to the user; another method It is a recommendation method based on short videos. Given a user, the recommendation result is given by recommending the most similar video to the user's browsing video. In addition, in the traditional recommendation process, the recommendation technology based on the matching degree can also calculate the matching degree between a given user and the candidate set of videos, and then sort the candidate set of videos according to the matching degree, and then recommend several videos with the highest matching degree. However, in the traditional short video recommendation process, the premise of getting accurate recommendations needs to accumulate enough user browsing click data, and when the amount of user browsing data is small, accurate short video recommendation cannot be made.

基于此，对于拥有大量的社交网络数据的平台，可以利用用户的社交网络数据进行短视频推荐。相比传统的推荐方法，由于用户的社交网络中含有大量具有相同兴趣的好友，此外，用户通常与好友有着相似的圈子与兴趣如：金融圈、学生圈、科研圈、消费等级、篮球、羽毛球、游泳等。基于社交网络中的社交关系进行短视频推荐，能够通过对与用户账号具有社交关系的账号的聚合来描述用户账号，从而更好的通过社交信息进行短视频推荐，提高了短视频推荐的准确性。Based on this, for platforms with a large amount of social network data, users' social network data can be used to recommend short videos. Compared with the traditional recommendation method, since the user's social network contains a large number of friends with the same interests, in addition, the user usually has similar circles and interests with friends, such as: financial circle, student circle, scientific research circle, consumption level, basketball, badminton , swimming, etc. The short video recommendation based on the social relationship in the social network can describe the user account through the aggregation of accounts that have a social relationship with the user account, so as to better recommend short videos through social information and improve the accuracy of short video recommendations. .

具体地，对于社交网络G＝(A,X)，A是账号交互特征，根据社交网络中各用户的社交账号之间的交互行为得到，A具体可以用户网络矩阵，若社交网络中的全体用户数量为N，则用户网络矩阵A中包括N各用户的社交账号之间的交互特征。用户网络矩阵A根据多种交互行为信息构建，如根据用户聊天数量、用户聊天频率、用户朋友圈互动评率等交互行为特征生成。X为用户的社交账号的账号特征。对于社交网络G，可以基于用户历史行为，如用户之间的交互行为，将全网用户连接成用户网络，用户间的连边由用户亲密度决定，用户亲密度根据用户之间的交互行为确定。如果两个用户间有较多交互行为，则两个用户连线权重较高，如果两个用户间交互行为较少，则两个用户连线权重较低。如果两个用户间没有交互行为，则两个用户间没有连线。其中衡量用户交互行为可以通过用户交互行为次数、交互行为累计时长、交互人数、交互频率排序、最近一星期交互天数等各种指标共同决定。Specifically, for a social network G=(A,X), A is the account interaction feature, which is obtained according to the interaction behavior between the social accounts of each user in the social network, A can specifically be a user network matrix, if all users in the social network The number is N, then the user network matrix A includes the interaction features between the social accounts of N users. The user network matrix A is constructed based on a variety of interactive behavior information, such as the number of user chats, the frequency of user chats, the interaction rate of user circles of friends and other interactive behavior characteristics. X is an account feature of the user's social account. For the social network G, based on the historical behavior of users, such as the interaction between users, the users of the entire network can be connected into a user network. The connection between users is determined by the user intimacy, and the user intimacy is determined by the interaction between users. . If there are more interactions between two users, the weight of the connection between the two users is higher, and if there is less interaction between the two users, the weight of the connection between the two users is lower. If there is no interaction between two users, there is no connection between the two users. Among them, the measurement of user interaction behavior can be jointly determined by various indicators such as the number of user interaction behaviors, the cumulative duration of interaction behaviors, the number of people interacting, the ranking of interaction frequencies, and the number of days of interaction in the last week.

例如，对于用户u1、u2和u3，若u1、u2交互次数较多，则u1与u2之间的连线权重较高；若u1和u3的交互次数较少，则u1与u3的连线权重较低；若u1与其他用户无交互，则u1与其他用户无连线。进一步地，对于用户i，j之间的亲密度，可以令用户i和用户j之间的交互次数为c_ij，而用户i和用户j之间的交互特征可以为A_ij＝log(1+c_ij)，用户网络矩阵A根据社交网络中所有用户之间的交互特征得到。For example, for users u1, u2 and u3, if u1 and u2 have more interactions, the connection weight between u1 and u2 is higher; if u1 and u3 have fewer interactions, the connection weight between u1 and u3 Low; if u1 has no interaction with other users, then u1 has no connection with other users. Further, for the intimacy between users i and j, the number of interactions between user i and j can be c_ij , and the interaction feature between user i and j can be A_ij =log(1+ c_ij ), the user network matrix A is obtained according to the interaction characteristics between all users in the social network.

进一步地，用户的社交账号的账号特征X通过对用户的社交账号进行向量化嵌入处理得到。具体地，通过向量特征来描述社交网络中的每一个用户，向量化嵌入的物理意义是通过向量来描述用户的社交行为，使得社交关系相近的用户的向量表示比较相近，相应的社交关系比较遥远的用户，向量表示差异较大。具体采用Node2Vec节点嵌入的无监督用户嵌入方法，对社交网络中的用户进行嵌入处理，如可以以每个用户对应的社交账号为节点，从每个节点出发，随机游走多条轨迹，并将全部游走出的轨迹作为语料库输入至word2vec词向量嵌入模型中进行向量化嵌入，得到用户的社交账号的账号特征X。其中，word2vec词向量嵌入模型可以基于word2vec算法进行训练，每个用户都可以通过一个向量进行表示。此外，社交网络中相近的节点表示较为相似，同时，由于图中不同用户之间的连线权重不同，因此在进行嵌入时考虑到权重的影响使用带权随机游走，得到用户的社交账号的账号特征。具体地，X＝{x₁,x₂,...x_N}，其中，x₁,x₂,...x_N为每个用户的社交账号的账号特征，X为社交网络中所有用户的社交账号的账号特征构成的矩阵，N为社交网络中用户数量。Further, the account feature X of the user's social account is obtained by vectorizing and embedding the user's social account. Specifically, each user in the social network is described by vector features. The physical meaning of vectorized embedding is to describe the social behavior of users through vectors, so that the vector representations of users with similar social relationships are relatively similar, and the corresponding social relationships are relatively distant. users, the vector representations are quite different. Specifically, the unsupervised user embedding method of Node2Vec node embedding is used to embed users in the social network. For example, the social account corresponding to each user can be used as a node, starting from each node, randomly walking multiple trajectories, and All the trajectories traveled out are input into the word2vec word vector embedding model as a corpus for vectorized embedding, and the account feature X of the user's social account is obtained. Among them, the word2vec word vector embedding model can be trained based on the word2vec algorithm, and each user can be represented by a vector. In addition, similar nodes in the social network are more similar. At the same time, because the connection weights between different users in the graph are different, the weighted random walk is used to obtain the user's social account number when embedding. Account characteristics. Specifically, X={x₁ , x₂ ,...x_N }, where x₁ , x₂ ,...x_N is the account feature of each user's social account, and X is all users in the social network is a matrix formed by the account features of social accounts, and N is the number of users in the social network.

进一步地，通过对访问某一短视频的所有用户的特征进行聚合，从而使得可以通过向量化的方式来表示每个短视频。每个短视频对应的向量的物理意义为，每个短视频的向量表示与喜欢访问或浏览该短视频的用户的特征点乘，即用户与该短视频的相似度高。具体地，对于某一短视频，访问过该短视频的用户列表为(u1,u2,u3,u4,…uK)，K为访问过该短视频的所有用户的总量。则可以通过对访问过该短视频的用户的账号特征进行平均池化(mean pooling)得到，即短视频的信息交互特征

对应的，所有短视频的信息交互特征构成的矩阵I＝{i₁,i₂,...i_n}，n为所有短视频的数量。Furthermore, by aggregating the features of all users who visit a certain short video, each short video can be expressed in a vectorized manner. The physical meaning of the vector corresponding to each short video is that the vector representation of each short video is multiplied by the feature point of the user who likes to visit or browse the short video, that is, the similarity between the user and the short video is high. Specifically, for a certain short video, the list of users who have visited the short video is (u1, u2, u3, u4,...uK), and K is the total number of all users who have visited the short video. Then it can be obtained by performing mean pooling on the account features of users who have visited the short video, that is, the information interaction characteristics of the short video

Correspondingly, the matrix I={i₁ , i₂ ,...i_n } formed by the information interaction features of all short videos, where n is the number of all short videos.

进一步地，通过社交关系聚合的方式，更好的描述每个用户的信息。通过基于社交关系聚合用户特征，可以在用户自身特征的基础上，同时考虑用户对应的交互用户的信息，通过对用户邻居的描述，可以更好的描述该用户的性质和兴趣，从而提高短视频推荐的准确性。具体地，利用用户的社交账号矩阵X以及用户网络矩阵A来进行信息传递。为了保证多次特征传播映射之后各用户的信息量的信息能保持在固定的量级上，可以对用户网络矩阵A进行标准化处理，具体可以通过下列三式中的任一公式对用户网络矩阵A进行标准化处理：Further, the information of each user is better described through social relationship aggregation. By aggregating user features based on social relations, it is possible to consider the information of the interactive user corresponding to the user on the basis of the user's own characteristics. By describing the user's neighbors, the user's nature and interests can be better described, thereby improving short video. Recommended accuracy. Specifically, the user's social account matrix X and user network matrix A are used for information transfer. In order to ensure that the amount of information of each user can be maintained at a fixed level after multiple feature propagation mappings, the user network matrix A can be standardized. Specifically, the user network matrix A can be calculated by any of the following three formulas To normalize:

其中，

为用户网络矩阵A进行标准化处理后的结果，D为对角矩阵，具体D_ij＝∑_jA_ij，I为单位矩阵。in,

is the result of standardization processing of the user network matrix A, D is a diagonal matrix, specifically D_ij =∑_j A_ij , and I is an identity matrix.

在对所有用户的社交账号的账号特征构成的矩阵X进行特征传播映射时，未进行特征传播映射的用户特征H⁽⁰⁾＝X，通过下式进行特征传播映射，When carrying out feature propagation mapping to the matrix X formed by the account features of all users' social accounts, the user feature H⁽⁰⁾ =X without feature propagation mapping is carried out by the following formula for feature propagation mapping,

其中，H^(l+1)为第一次特征传播映射的结果，σ为非线性函数，如可以为RELU函数，H^(l)为第一次特征传播映射的初始值，W^(l)为第一次特征传播映射的映射矩阵。Among them, H^(l+1) is the result of the first feature propagation mapping, σ is a nonlinear function, such as RELU function, H^(l) is the initial value of the first feature propagation mapping, W^(l) is The mapping matrix for the first feature propagation mapping.

若特征传播映射的传播次数为K，则可以得到，If the propagation times of the feature propagation map is K, it can be obtained,

其中，H^(K)为K次特征传播映射的结果。Among them, H^(K) is the result of K times feature propagation mapping.

通过社交网络中的社交关系进行特征传播映射后，可以通过多层感知机来预测用户和短视频的匹配度，即可以将用户u和视频i的表示输入到多层感知机中得到预测结果。如针对用户u和视频i，用户短视频对ui的预测匹配度为如下式，After the feature propagation mapping is performed through the social relationship in the social network, the matching degree between the user and the short video can be predicted through the multi-layer perceptron, that is, the representation of user u and video i can be input into the multi-layer perceptron to obtain the prediction result. For example, for user u and video i, the predicted matching degree of the user's short video to ui is as follows,

其中，

为用户u与短视频i的匹配度，MLP()为多层感知机，U_u为用户u的账号特征，I_i为短视频i的特征。在训练多层感知机时，可以通过随机筛选正样本(即真实发生的用户视频对)以及负样本(即未发生的用户视频对)来对MLP以及特征传播映射中的映射矩阵参数进行训练。in,

is the matching degree between user u and short video i, MLP() is a multi-layer perceptron, U_u is the account feature of user u, and I_i is the feature of short video i. When training a multi-layer perceptron, the MLP and the mapping matrix parameters in the feature propagation map can be trained by randomly selecting positive samples (that is, user video pairs that actually occurred) and negative samples (that is, user video pairs that did not occur).

进一步地，在针对目标用户进行短视频推荐时，可以先获取到社交网络中全部用户的网络账号特征U，然后当用户u触发短视频浏览请求时，通过上述短视频信息推荐方法，比较用户u与候选集中全部短视频的相似度，随后选择排序最高的一部分短视频进行推荐，从而利用社交网络中的社交关系进行短视频推荐，提高了短视频推荐的准确性。Further, when recommending short videos for target users, the network account features U of all users in the social network can be obtained first, and then when user u triggers a short video browsing request, the above short video information recommendation method is used to compare user The similarity between u and all short videos in the candidate set, and then select a part of the short videos with the highest ranking for recommendation, so as to use the social relationship in the social network to recommend short videos and improve the accuracy of short video recommendations.

本申请还另外提供一种应用场景，该应用场景应用上述的信息推荐方法。The present application further provides an application scenario, where the above-mentioned information recommendation method is applied.

具体地，该信息推荐方法在该应用场景的应用如下：Specifically, the application of the information recommendation method in this application scenario is as follows:

用户通过社交网络中的社交账号登录电子书平台，在用户触发进行电子书推荐时，服务器获取用户账号所属社交网络对应的账号交互特征，通过账号交互特征，对用户账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得用户账号对应的传播账号特征，服务器获取电子书平台中每个电子书各自的信息交互特征，并将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各电子书中确定的目标电子书向用户账号对应的终端进行推荐。A user logs in to the e-book platform through a social account in a social network. When the user triggers an e-book recommendation, the server obtains the account interaction characteristics corresponding to the social network to which the user account belongs, and uses the account interaction characteristics to perform feature propagation on the account characteristics corresponding to the user account. Mapping, according to the result of feature propagation mapping to obtain the characteristics of the communication account corresponding to the user account, the server obtains the respective information interaction characteristics of each e-book in the e-book platform, and matches the characteristics of the propagation account with each information interaction characteristic, and based on the matching As a result, the target e-books determined in each e-book are recommended to the terminal corresponding to the user account.

此外，在其他应用场景中，信息推荐所针对的还可以为音乐、图片、网页等各种互联网信息资源。In addition, in other application scenarios, the information recommendation may target various Internet information resources such as music, pictures, and web pages.

应该理解的是，虽然图2、图3、图5、图6和图9的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2、图3、图5、图6和图9中的至少一部分步骤可以包括多个步骤或者多个阶段，这些步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤中的步骤或者阶段的至少一部分轮流或者交替地执行。It should be understood that although the various steps in the flow charts of FIG. 2 , FIG. 3 , FIG. 5 , FIG. 6 and FIG. 9 are shown sequentially as indicated by the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in FIG. 2, FIG. 3, FIG. 5, FIG. 6 and FIG. The execution sequence of these steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with other steps or at least a part of steps or stages in other steps.

在一个实施例中，如图10所示，提供了一种信息推荐装置1000，该装置可以采用软件模块或硬件模块，或者是二者的结合成为计算机设备的一部分，该装置具体包括：账号交互特征获取模块1002、特征传播映射模块1004、信息交互特征获取模块1006和目标信息推荐模块1008，其中：In one embodiment, as shown in FIG. 10 , an information recommendation apparatus 1000 is provided. The apparatus may adopt software modules or hardware modules, or a combination of the two to become a part of computer equipment. The apparatus specifically includes: account interactionFeature acquisition module 1002, featurepropagation mapping module 1004, information interactionfeature acquisition module 1006 and targetinformation recommendation module 1008, wherein:

账号交互特征获取模块1002，用于获取目标账号所属社交网络对应的账号交互特征；账号交互特征根据社交网络中各社交账号之间的交互行为得到；The account interactionfeature acquisition module 1002 is used to acquire the account interaction feature corresponding to the social network to which the target account belongs; the account interaction feature is obtained according to the interaction behavior between social accounts in the social network;

特征传播映射模块1004，用于通过账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征；The featurepropagation mapping module 1004 is used to perform feature propagation mapping on the account characteristics corresponding to the target account through the account interaction characteristics, and obtain the propagation account characteristics corresponding to the target account according to the result of the feature propagation mapping;

信息交互特征获取模块1006，用于获取每个待推荐信息各自的信息交互特征；信息交互特征是基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；The information interactionfeature acquisition module 1006 is used to acquire the respective information interaction features of each information to be recommended; the information interaction features are generated based on the account features of social accounts that have interacted with the corresponding information to be recommended in the social network;

目标信息推荐模块1008，用于将传播账号特征和各信息交互特征进行匹配，并将根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。The targetinformation recommending module 1008 is configured to match the characteristics of the dissemination account and each information interaction feature, and recommend the target information determined from each information to be recommended according to the matching result to the terminal corresponding to the target account.

上述信息推荐装置，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得目标账号对应的传播账号特征，并将传播账号特征与每个待推荐信息各自的信息交互特征进行匹配，信息交互特征基于社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成，根据匹配结果从各待推荐信息中确定的目标信息向目标账号对应的终端进行推荐。在信息推荐处理过程中，通过根据社交网络中各社交账号之间的交互行为得到的账号交互特征，对目标账号对应的账号特征进行特征传播映射，从而将社交网络中与目标账号具有交互行为关系的社交账号的账号特征传播映射至目标账号，得到目标账号对应的传播账号特征，并通过传播账号特征和待推荐信息各自的信息交互特征进行匹配，根据匹配结果确定目标信息进行推荐，从而利用了社交网络中各社交账号之间的交互行为关系进行信息推荐，提高了信息推荐的准确性。The above information recommendation device performs feature propagation mapping on the account features corresponding to the target account through the account interaction features obtained according to the interaction behavior between social accounts in the social network, and obtains the propagation account features corresponding to the target account according to the result of feature propagation mapping. , and match the dissemination account characteristics with the information interaction characteristics of each information to be recommended. The information interaction characteristics are generated based on the account characteristics of the social accounts that have interacted with the corresponding information to be recommended in the social network. The target information determined in the recommendation information is recommended to the terminal corresponding to the target account. In the process of information recommendation processing, through the account interaction features obtained according to the interaction behavior between social accounts in the social network, the account features corresponding to the target account are subjected to feature propagation mapping, so that the social network has an interactive behavior relationship with the target account. The account features of the social account are propagated and mapped to the target account, and the corresponding spread account features of the target account are obtained, and the spread account features are matched with the information interaction features of the information to be recommended, and the target information is determined according to the matching results for recommendation, thus using Information recommendation is performed based on the interactive behavior relationship between social accounts in the social network, which improves the accuracy of information recommendation.

在一个实施例中，特征传播映射模块1004包括目标交互特征提取模块、交互账号特征获取模块、传播映射迭代模块和传播账号特征确定模块；其中：目标交互特征提取模块，用于从账号交互特征中提取与目标账号对应的目标账号交互特征；交互账号特征获取模块，用于获取目标账号交互特征对应的交互社交账号的账号特征；传播映射迭代模块，用于基于目标账号交互特征、交互社交账号的账号特征和传播映射参数，对目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果；传播账号特征确定模块，用于根据特征传播映射的结果确定目标账号对应的传播账号特征。In one embodiment, the featurepropagation mapping module 1004 includes a target interaction feature extraction module, an interaction account feature acquisition module, a propagation mapping iteration module, and a propagation account feature determination module; wherein: the target interaction feature extraction module is used to extract from the account interaction feature Extract the target account interaction characteristics corresponding to the target account; the interactive account feature acquisition module is used to obtain the account characteristics of the interactive social accounts corresponding to the target account interaction characteristics; The account features and propagation mapping parameters are used to perform iterative feature propagation mapping on the account characteristics corresponding to the target account to obtain the result of the feature propagation mapping; the propagation account feature determination module is used to determine the propagation account characteristics corresponding to the target account according to the result of the feature propagation mapping.

在一个实施例中，传播映射迭代模块包括当前账号特征确定模块和传播映射处理模块；其中：当前账号特征确定模块，用于将目标账号对应的账号特征确定为当前账号特征；传播映射处理模块，用于基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果；将本次迭代特征传播映射的结果作为当前账号特征，并返回基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果的步骤。In one embodiment, the propagation mapping iteration module includes a current account feature determination module and a propagation mapping processing module; wherein: the current account feature determination module is used to determine the account feature corresponding to the target account as the current account feature; the propagation mapping processing module, It is used to perform feature propagation mapping on the characteristics of the current account based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping to obtain the result of this iterative feature propagation mapping; The result of the propagation mapping is used as the current account feature, and returns the propagation mapping parameters based on the interaction characteristics of the target account, the account features of the interactive social account, and the feature propagation mapping of this iteration, and performs feature propagation mapping on the characteristics of the current account to obtain the feature propagation of this iteration Steps to map the results.

在一个实施例中，传播映射处理模块包括特征传播模块和非线性映射模块；其中：特征传播模块，用于基于目标账号交互特征、交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对当前账号特征进行特征传播，得到特征传播结果；非线性映射模块，用于对特征传播结果进行非线性映射，得到本次迭代特征传播映射的结果。In one embodiment, the propagation mapping processing module includes a feature propagation module and a nonlinear mapping module; wherein: the feature propagation module is used for the propagation mapping based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the iterative feature propagation mapping parameter, perform feature propagation on the characteristics of the current account, and obtain the result of feature propagation; the nonlinear mapping module is used to perform nonlinear mapping on the result of feature propagation, and obtain the result of this iterative feature propagation mapping.

在一个实施例中，特征传播映射模块1004包括网络账号特征确定模块、模型特征传播映射模块和模型输出处理模块；其中：网络账号特征确定模块，用于确定根据社交网络中各社交账号的账号特征构建的网络账号特征；网络账号特征包括目标账号对应的账号特征；模型特征传播映射模块，用于将账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射，获得特征传播映射模型输出的网络传播账号特征；模型输出处理模块，用于从网络传播账号特征中提取目标账号对应的传播账号特征。In one embodiment, the featurepropagation mapping module 1004 includes a network account feature determination module, a model feature propagation mapping module, and a model output processing module; The constructed network account features; the network account features include the account features corresponding to the target account; the model feature propagation mapping module is used to input the account interaction features and network account features into the feature propagation mapping model for feature propagation mapping, and obtain the output of the feature propagation mapping model The characteristics of the network communication account; the model output processing module is used to extract the characteristics of the communication account corresponding to the target account from the characteristics of the network communication account.

在一个实施例中，模型特征传播映射模块包括标准化处理模块和特征输入模块；其中：标准化处理模块，用于通过标准化条件对账号交互特征进行标准化处理，得到标准化的账号交互特征；特征输入模块，用于将标准化的账号交互特征和网络账号特征输入特征传播映射模型中进行特征传播映射。In one embodiment, the model feature propagation and mapping module includes a standardized processing module and a feature input module; wherein: the standardized processing module is used to standardize the account interaction features through standardized conditions to obtain standardized account interaction features; the feature input module, It is used to input the standardized account interaction features and network account features into the feature propagation mapping model for feature propagation mapping.

在一个实施例中，目标信息推荐模块1008包括特征匹配模块、目标信息确定模块和目标信息处理模块；其中：特征匹配模块，用于通过匹配模型对传播账号特征和各信息交互特征进行匹配，得到匹配模型输出的匹配结果；目标信息确定模块，用于基于匹配结果从各待推荐信息中确定目标信息；目标信息处理模块，用于将目标信息向目标账号对应的终端进行推荐。In one embodiment, the targetinformation recommendation module 1008 includes a feature matching module, a target information determination module, and a target information processing module; wherein: the feature matching module is used to match the characteristics of the dissemination account and each information interaction feature through a matching model to obtain The matching result output by the matching model; the target information determining module, configured to determine the target information from each information to be recommended based on the matching result; the target information processing module, configured to recommend the target information to the terminal corresponding to the target account.

在一个实施例中，目标信息确定模块包括推荐条件筛选模块和筛选结果处理模块；其中：推荐条件筛选模块，用于从匹配结果中确定满足推荐条件的目标匹配结果；筛选结果处理模块，用于将目标匹配结果所对应的待推荐信息确定为目标信息。In one embodiment, the target information determination module includes a recommendation condition screening module and a screening result processing module; wherein: the recommendation condition screening module is used to determine the target matching result satisfying the recommendation condition from the matching results; the screening result processing module is used to The information to be recommended corresponding to the target matching result is determined as the target information.

在一个实施例中，还包括统计结果确定模块、交互特征获得模块和账号交互特征生成模块；其中：统计结果确定模块，用于确定目标账号所属的社交网络中各社交账号之间交互行为的统计结果；交互特征获得模块，用于基于统计结果获得社交网络中各社交账号之间的交互特征；账号交互特征生成模块，用于根据社交网络中各社交账号之间的交互特征，生成社交网络对应的账号交互特征。In one embodiment, it also includes a statistical result determination module, an interaction feature acquisition module, and an account interaction feature generation module; wherein: the statistical result determination module is used to determine the statistics of the interaction behavior between social accounts in the social network to which the target account belongs Result; the interaction feature acquisition module is used to obtain the interaction features between the social accounts in the social network based on statistical results; the account interaction feature generation module is used to generate social network correspondence according to the interaction features between the social accounts in the social network account interaction features.

在一个实施例中，还包括社交网络确定模块、节点嵌入模块和账号特征确定模块；其中：社交网络确定模块，用于确定目标账号所属的社交网络；节点嵌入模块，用于对社交网络中的各节点进行节点嵌入，得到各节点分别对应的节点特征；各节点与社交网络中的各社交账号对应，各节点之间的节点关系与各社交账号之间的交互行为对应；账号特征确定模块，用于基于各节点对应的节点特征，确定目标账号对应的账号特征。In one embodiment, it also includes a social network determination module, a node embedding module and an account feature determination module; wherein: a social network determination module is used to determine the social network to which the target account belongs; a node embedding module is used to identify the social network in the social network Node embedding is performed on each node to obtain node characteristics corresponding to each node; each node corresponds to each social account in the social network, and the node relationship between each node corresponds to the interactive behavior between each social account; the account feature determination module, It is used to determine the account characteristics corresponding to the target account based on the node characteristics corresponding to each node.

在一个实施例中，节点嵌入模块包括权值确定模块、游走模块和游走轨迹处理模块；其中：权值确定模块，用于根据社交网络中各节点之间的节点关系，确定各节点之间的游走权重；游走模块，用于以每个节点为起点，基于游走权重在社交网络中进行节点游走，形成各节点游走轨迹；游走轨迹处理模块，用于通过嵌入模型对各节点游走轨迹进行特征嵌入，得到各节点分别对应的节点特征。In one embodiment, the node embedding module includes a weight determination module, a walking module, and a walking trajectory processing module; wherein: the weight determination module is used to determine the relationship between each node according to the node relationship between each node in the social network. The walking weight between them; the walking module is used to take each node as a starting point and perform node walking in the social network based on the walking weight to form the walking trajectory of each node; the walking trajectory processing module is used to pass the embedding model The feature embedding is performed on the walking trajectory of each node to obtain the node features corresponding to each node.

在一个实施例中，还包括交互社交账号确定模块和信息交互特征获得模块；其中：交互社交账号确定模块，用于确定每个待推荐信息对各自的交互社交账号；交互社交账号为社交网络中对相应待推荐信息产生过交互操作的社交账号；信息交互特征获得模块，用于对各交互社交账号对应的账号特征进行特征聚合，得到相应待推荐信息的信息交互特征。In one embodiment, it also includes an interactive social account number determination module and an information interaction feature acquisition module; wherein: the interactive social account number determination module is used to determine the respective interactive social account number of each information to be recommended; the interactive social account number is a social network The social accounts that have interacted with the corresponding information to be recommended; the information interaction feature acquisition module is used to aggregate the account features corresponding to each interactive social account to obtain the information interaction features of the corresponding information to be recommended.

关于信息推荐装置的具体限定可以参见上文中对于信息推荐方法的限定，在此不再赘述。上述信息推荐装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。For specific limitations on the information recommendation apparatus, refer to the above-mentioned limitations on the information recommendation method, which will not be repeated here. Each module in the above-mentioned information recommending device can be fully or partially realized by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can invoke and execute the corresponding operations of the above-mentioned modules.

在一个实施例中，提供了一种计算机设备，该计算机设备可以是服务器，其内部结构图可以如图11所示。该计算机设备包括通过系统总线连接的处理器、存储器和网络接口。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的数据库用于存储数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种信息推荐方法。In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure may be as shown in FIG. 11 . The computer device includes a processor, memory and a network interface connected by a system bus. Wherein, the processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs and databases. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, an information recommendation method is realized.

本领域技术人员可以理解，图11中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。Those skilled in the art can understand that the structure shown in Figure 11 is only a block diagram of a part of the structure related to the solution of this application, and does not constitute a limitation on the computer equipment on which the solution of this application is applied. The specific computer equipment can be More or fewer components than shown in the figures may be included, or some components may be combined, or have a different arrangement of components.

在一个实施例中，还提供了一种计算机设备，包括存储器和处理器，存储器中存储有计算机程序，该处理器执行计算机程序时实现上述各方法实施例中的步骤。In one embodiment, there is also provided a computer device, including a memory and a processor, where a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

在一个实施例中，提供了一种计算机可读存储介质，存储有计算机程序，该计算机程序被处理器执行时实现上述各方法实施例中的步骤。In one embodiment, a computer-readable storage medium is provided, storing a computer program, and implementing the steps in the foregoing method embodiments when the computer program is executed by a processor.

在一个实施例中，提供了一种计算机程序产品或计算机程序，该计算机程序产品或计算机程序包括计算机指令，该计算机指令存储在计算机可读存储介质中。计算机设备的处理器从计算机可读存储介质读取该计算机指令，处理器执行该计算机指令，使得该计算机设备执行上述各方法实施例中的步骤。In one embodiment there is provided a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the foregoing method embodiments.

本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和易失性存储器中的至少一种。非易失性存储器可包括只读存储器(Read-Only Memory，ROM)、磁带、软盘、闪存或光存储器等。易失性存储器可包括随机存取存储器(Random Access Memory，RAM)或外部高速缓冲存储器。作为说明而非局限，RAM可以是多种形式，比如静态随机存取存储器(Static Random Access Memory，SRAM)或动态随机存取存储器(Dynamic Random Access Memory，DRAM)等。Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware, and the computer programs can be stored in a non-volatile computer-readable memory In the medium, when the computer program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, any references to memory, storage, database or other media used in the various embodiments provided in the present application may include at least one of non-volatile memory and volatile memory. The non-volatile memory may include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory or optical memory, and the like. Volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。The technical features of the above embodiments can be combined arbitrarily. To make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be It is considered to be within the range described in this specification.

以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。The above-mentioned embodiments only represent several implementation modes of the present application, and the description thereof is relatively specific and detailed, but it should not be construed as limiting the scope of the patent for the invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the scope of protection of the patent application should be based on the appended claims.

Claims (15)

Translated fromChinese

1.一种信息推荐方法，其特征在于，所述方法包括：1. A method for information recommendation, characterized in that the method comprises:获取目标账号所属社交网络对应的账号交互特征；所述账号交互特征根据所述社交网络中各社交账号之间的交互行为得到；Obtaining account interaction features corresponding to the social network to which the target account belongs; the account interaction features are obtained according to interaction behaviors between social accounts in the social network;通过所述账号交互特征，对所述目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得所述目标账号对应的传播账号特征；Perform feature propagation mapping on the account features corresponding to the target account through the account interaction features, and obtain the propagation account features corresponding to the target account according to the result of the feature propagation mapping;获取每个待推荐信息各自的信息交互特征；所述信息交互特征是基于所述社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；Acquiring the respective information interaction characteristics of each information to be recommended; the information interaction characteristics are generated based on the account characteristics of social accounts that have interacted with the corresponding information to be recommended in the social network;将所述传播账号特征和各所述信息交互特征进行匹配，并将根据匹配结果从各所述待推荐信息中确定的目标信息向所述目标账号对应的终端进行推荐。Matching the characteristics of the dissemination account with each of the information interaction characteristics, and recommending the target information determined from each of the information to be recommended according to the matching result to the terminal corresponding to the target account.2.根据权利要求1所述的方法，其特征在于，所述通过所述账号交互特征，对所述目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得所述目标账号对应的传播账号特征，包括：2. The method according to claim 1, wherein the feature propagation mapping is performed on the account characteristics corresponding to the target account through the account interaction characteristics, and the target account corresponding to the target account is obtained according to the result of the feature propagation mapping. The characteristics of the dissemination account, including:从所述账号交互特征中提取与所述目标账号对应的目标账号交互特征；Extracting target account interaction features corresponding to the target account from the account interaction features;获取所述目标账号交互特征对应的交互社交账号的账号特征；Acquiring the account characteristics of the interactive social account corresponding to the target account interaction characteristics;基于所述目标账号交互特征、所述交互社交账号的账号特征和传播映射参数，对所述目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果；Based on the interaction characteristics of the target account, the account characteristics of the interactive social account and the propagation mapping parameters, iterative feature propagation mapping is performed on the account characteristics corresponding to the target account to obtain a result of feature propagation mapping;根据所述特征传播映射的结果确定所述目标账号对应的传播账号特征。The characteristics of the spreading account corresponding to the target account are determined according to the result of the characteristic spreading mapping.3.根据权利要求2所述的方法，其特征在于，所述基于所述目标账号交互特征、所述交互社交账号的账号特征和传播映射参数，对所述目标账号对应的账号特征进行迭代特征传播映射，得到特征传播映射的结果，包括：3. The method according to claim 2, wherein the iterative feature is performed on the account feature corresponding to the target account based on the target account interaction feature, the account feature of the interactive social account, and the propagation mapping parameter. Propagation map, get the result of feature propagation map, including:将所述目标账号对应的账号特征确定为当前账号特征；determining the account feature corresponding to the target account as the current account feature;基于所述目标账号交互特征、所述交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对所述当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果；Based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative feature propagation mapping, perform feature propagation mapping on the current account characteristics, and obtain the result of this iterative feature propagation mapping;将本次迭代特征传播映射的结果作为所述当前账号特征，并返回所述基于所述目标账号交互特征、所述交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对所述当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果的步骤。Using the result of this iterative feature propagation mapping as the current account feature, and returning the propagation mapping parameters based on the target account interaction feature, the account feature of the interactive social account, and this iterative feature propagation mapping, for all The step of performing feature propagation mapping on the characteristics of the current account to obtain the result of this iterative feature propagation mapping.4.根据权利要求3所述的方法，其特征在于，所述基于所述目标账号交互特征、所述交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对所述当前账号特征进行特征传播映射，得到本次迭代特征传播映射的结果，包括：4. The method according to claim 3, characterized in that, based on the target account interaction characteristics, the account characteristics of the interactive social account and the propagation mapping parameters of this iterative characteristic propagation mapping, the current account Features are subjected to feature propagation mapping to obtain the results of this iterative feature propagation mapping, including:基于所述目标账号交互特征、所述交互社交账号的账号特征和本次迭代特征传播映射的传播映射参数，对所述当前账号特征进行特征传播，得到特征传播结果；Based on the interaction characteristics of the target account, the account characteristics of the interactive social account, and the propagation mapping parameters of this iterative characteristic propagation mapping, perform feature propagation on the current account characteristics to obtain a feature propagation result;对所述特征传播结果进行非线性映射，得到本次迭代特征传播映射的结果。A non-linear mapping is performed on the feature propagation result to obtain the result of the iterative feature propagation mapping.5.根据权利要求1所述的方法，其特征在于，所述通过所述账号交互特征，对所述目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得所述目标账号对应的传播账号特征，包括：5. The method according to claim 1, wherein the feature propagation mapping is performed on the account characteristics corresponding to the target account through the account interaction characteristics, and the target account corresponding to the target account is obtained according to the result of the feature propagation mapping. The characteristics of the dissemination account, including:确定根据所述社交网络中各社交账号的账号特征构建的网络账号特征；所述网络账号特征包括所述目标账号对应的账号特征；Determining network account features constructed according to the account features of each social account in the social network; the network account features include account features corresponding to the target account;将所述账号交互特征和所述网络账号特征输入特征传播映射模型中进行特征传播映射，获得所述特征传播映射模型输出的网络传播账号特征；Inputting the account interaction features and the network account features into a feature propagation mapping model to perform feature propagation mapping, and obtain the network propagation account characteristics output by the feature propagation mapping model;从所述网络传播账号特征中提取所述目标账号对应的传播账号特征。The characteristics of the communication account corresponding to the target account are extracted from the characteristics of the network communication account.6.根据权利要求5所述的方法，其特征在于，所述将所述账号交互特征和所述网络账号特征输入特征传播映射模型中进行特征传播映射，包括：6. The method according to claim 5, wherein said inputting said account interaction features and said network account features into a feature propagation mapping model to perform feature propagation mapping comprises:通过标准化条件对所述账号交互特征进行标准化处理，得到标准化的账号交互特征；Standardize the account interaction features through standardization conditions to obtain standardized account interaction features;将所述标准化的账号交互特征和所述网络账号特征输入特征传播映射模型中进行特征传播映射。Inputting the standardized account interaction features and the network account features into a feature propagation mapping model to perform feature propagation mapping.7.根据权利要求1所述的方法，其特征在于，所述将所述传播账号特征和各所述信息交互特征进行匹配，并将根据匹配结果从各所述待推荐信息中确定的目标信息向所述目标账号对应的终端进行推荐，包括：7. The method according to claim 1, characterized in that, the characteristics of the dissemination account are matched with each of the information interaction characteristics, and the target information determined from each of the information to be recommended is selected according to the matching result Recommending to the terminal corresponding to the target account, including:通过匹配模型对所述传播账号特征和各所述信息交互特征进行匹配，得到所述匹配模型输出的匹配结果；Matching the characteristics of the dissemination account and each of the information interaction characteristics through a matching model to obtain a matching result output by the matching model;基于所述匹配结果从各所述待推荐信息中确定目标信息；determining target information from each of the information to be recommended based on the matching result;将所述目标信息向所述目标账号对应的终端进行推荐。Recommending the target information to a terminal corresponding to the target account.8.根据权利要求7所述的方法，其特征在于，所述基于所述匹配结果从各所述待推荐信息中确定目标信息包括：8. The method according to claim 7, wherein said determining target information from each of said information to be recommended based on said matching result comprises:从所述匹配结果中确定满足推荐条件的目标匹配结果；Determining a target matching result that satisfies the recommendation condition from the matching results;将所述目标匹配结果所对应的待推荐信息确定为目标信息。The information to be recommended corresponding to the target matching result is determined as the target information.9.根据权利要求1所述的方法，其特征在于，所述方法还包括：9. The method according to claim 1, further comprising:确定目标账号所属的社交网络中各社交账号之间交互行为的统计结果；Determine the statistical results of the interaction between social accounts in the social network to which the target account belongs;基于所述统计结果获得所述社交网络中各社交账号之间的交互特征；Obtaining interaction features between social accounts in the social network based on the statistical results;根据所述社交网络中各社交账号之间的交互特征，生成所述社交网络对应的账号交互特征。According to the interaction features between social accounts in the social network, account interaction features corresponding to the social network are generated.10.根据权利要求1至9任意一项所述的方法，其特征在于，所述方法还包括：10. The method according to any one of claims 1 to 9, wherein the method further comprises:确定目标账号所属的社交网络；Determine the social network to which the target account belongs;对所述社交网络中的各节点进行节点嵌入，得到各节点分别对应的节点特征；各所述节点与所述社交网络中的各社交账号对应，各所述节点之间的节点关系与各所述社交账号之间的交互行为对应；Node embedding is performed on each node in the social network to obtain node characteristics corresponding to each node; each node corresponds to each social account in the social network, and the node relationship between each node is related to each node. Correspondence between interactive behaviors between social accounts;基于各所述节点对应的节点特征，确定所述目标账号对应的账号特征。Based on the node features corresponding to each of the nodes, the account features corresponding to the target account are determined.11.根据权利要求10所述的方法，其特征在于，所述对所述社交网络中的各节点进行节点嵌入，得到各节点分别对应的节点特征，包括：11. The method according to claim 10, wherein said carrying out node embedding to each node in the social network to obtain node features respectively corresponding to each node, comprising:根据所述社交网络中各节点之间的节点关系，确定各所述节点之间的游走权重；According to the node relationship between each node in the social network, determine the walking weight between each node;以每个所述节点为起点，基于所述游走权重在所述社交网络中进行节点游走，形成各节点游走轨迹；Taking each of the nodes as a starting point, performing node walking in the social network based on the walking weight to form a walking trajectory of each node;通过嵌入模型对各所述节点游走轨迹进行特征嵌入，得到各所述节点分别对应的节点特征。The feature embedding is performed on the walking trajectories of each of the nodes through the embedding model to obtain the node features corresponding to each of the nodes.12.根据权利要求1至11任意一项所述的方法，其特征在于，所述方法还包括：12. The method according to any one of claims 1 to 11, further comprising:确定每个待推荐信息对各自的交互社交账号；所述交互社交账号为所述社交网络中对相应待推荐信息产生过交互操作的社交账号；Determining each interactive social account for each information to be recommended; the interactive social account is a social account that has interacted with the corresponding information to be recommended in the social network;对各所述交互社交账号对应的账号特征进行特征聚合，得到相应待推荐信息的信息交互特征。The feature aggregation is performed on the account features corresponding to each of the interactive social accounts to obtain the information interaction features of the corresponding information to be recommended.13.一种信息推荐装置，其特征在于，所述装置包括：13. An information recommendation device, characterized in that the device comprises:账号交互特征获取模块，用于获取目标账号所属社交网络对应的账号交互特征；所述账号交互特征根据所述社交网络中各社交账号之间的交互行为得到；An account interaction feature acquisition module, configured to acquire the account interaction features corresponding to the social network to which the target account belongs; the account interaction features are obtained according to the interaction behavior between social accounts in the social network;特征传播映射模块，用于通过所述账号交互特征，对所述目标账号对应的账号特征进行特征传播映射，根据特征传播映射的结果获得所述目标账号对应的传播账号特征；A feature propagation mapping module, configured to perform feature propagation mapping on the account characteristics corresponding to the target account through the account interaction characteristics, and obtain the propagation account characteristics corresponding to the target account according to the result of the feature propagation mapping;信息交互特征获取模块，用于获取每个待推荐信息各自的信息交互特征；所述信息交互特征是基于所述社交网络中对相应待推荐信息产生过交互操作的社交账号的账号特征生成的；An information interaction feature acquisition module, configured to acquire the respective information interaction features of each information to be recommended; the information interaction features are generated based on the account features of social accounts that have interacted with the corresponding information to be recommended in the social network;目标信息推荐模块，用于将所述传播账号特征和各所述信息交互特征进行匹配，并将根据匹配结果从各所述待推荐信息中确定的目标信息向所述目标账号对应的终端进行推荐。A target information recommending module, configured to match the characteristics of the dissemination account with each of the information interaction features, and recommend the target information determined from each of the information to be recommended according to the matching result to the terminal corresponding to the target account .14.一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其特征在于，所述处理器执行所述计算机程序时实现权利要求1至12中任一项所述的方法的步骤。14. A computer device, comprising a memory and a processor, the memory stores a computer program, wherein the processor implements the method according to any one of claims 1 to 12 when executing the computer program step.15.一种计算机可读存储介质，存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现权利要求1至12中任一项所述的方法的步骤。15. A computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images