技术领域Technical Field
本申请的实施例涉及自媒体数据处理技术，尤其涉及一种自媒体用户选择方法及相关设备。The embodiments of the present application relate to self-media data processing technology, and more particularly to a self-media user selection method and related equipment.
背景技术Background technique
在相关的自媒体用户的筛选中，往往存在巨大的挑战，难以综合多方标准优选出影响力较大并有一定信息传播能力的目标用户，商家与自媒体用户的合作中，由于目标用户的选取难以准确，往往造成一定成本的浪费。There are often huge challenges in the screening of relevant self-media users. It is difficult to comprehensively consider multiple criteria to select target users with greater influence and certain information dissemination capabilities. In the cooperation between businesses and self-media users, it is often difficult to accurately select target users, which often results in a certain waste of costs.
基于此，需要一种能够实现综合多方面评价，准确挑选影响力较大的用户的方案。Based on this, a solution is needed that can achieve comprehensive multi-faceted evaluation and accurately select users with greater influence.
发明内容Summary of the invention
有鉴于此，本申请的目的在于提出一种自媒体用户选择方法及相关设备。In view of this, the purpose of this application is to propose a self-media user selection method and related equipment.
基于上述目的，本申请提供了自媒体用户选择方法，包括：Based on the above purpose, this application provides a self-media user selection method, including:
在自媒体平台中周期性获取多个预选用户的初始数据，通过对所述初始数据进行预处理，得到每个所述预选用户的关系特征、影响力特征、粉丝特征、情感特征和传播度特征；Periodically obtain initial data of multiple pre-selected users in the self-media platform, and obtain relationship characteristics, influence characteristics, fan characteristics, emotion characteristics, and communication characteristics of each pre-selected user by pre-processing the initial data;
利用所述关系特征，对全部所述预选用户执行聚类算法进行聚类，得到划分为不同类别的所述预选用户，在每个所述类别的所述预选用户中，按照预设比例确定多个候选用户；Using the relationship features, clustering the pre-selected users using a clustering algorithm to obtain the pre-selected users divided into different categories, and determining a plurality of candidate users in each category of the pre-selected users according to a preset ratio;
对全部所述候选用户中的每一个，采取极端梯度提升算法对该候选用户的所述影响力特征进行加权，得到影响力得分；利用所述粉丝特征计算粉丝重要度得分；利用自回归语言模型、循环卷积神经网络和注意力机制构建情感分析模型，对所述情感特征进行概率分析，得到情感得分；For each of the candidate users, the influence feature of the candidate user is weighted by using the extreme gradient boosting algorithm to obtain an influence score; the fan importance score is calculated using the fan feature; a sentiment analysis model is constructed using an autoregressive language model, a recurrent convolutional neural network and an attention mechanism, and a probability analysis is performed on the sentiment feature to obtain a sentiment score;
基于全部所述候选用户，利用所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分进行区间模糊处理，得到目标权重，将每个所述候选用户的所述目标权重输入预置的多准则决策框架，得到该候选用户的综合值，并根据所述综合值确定目标用户。Based on all the candidate users, interval fuzzy processing is performed using the spreadability characteristics, the influence score, the fan importance score and the sentiment score to obtain a target weight, and the target weight of each candidate user is input into a preset multi-criteria decision framework to obtain a comprehensive value of the candidate user, and the target user is determined based on the comprehensive value.
进一步地，所述在自媒体平台中周期性获取多个预选用户的初始数据，通过对所述初始数据进行预处理，得到每个所述预选用户的关系特征、影响力特征、粉丝特征、情感特征和传播度特征，包括：Furthermore, the initial data of multiple pre-selected users are periodically obtained in the self-media platform, and the relationship characteristics, influence characteristics, fan characteristics, emotion characteristics and communication characteristics of each pre-selected user are obtained by pre-processing the initial data, including:
在所述获取周期内，获取每个所述预选用户的初始数据；During the acquisition period, initial data of each of the pre-selected users is acquired;
提取该预选用户初始数据中的所述关系特征，所述关系特征包括与该预选用户互相关注的其他预选用户之间的关系数据；Extracting the relationship features from the initial data of the pre-selected user, wherein the relationship features include relationship data between other pre-selected users who mutually follow the pre-selected user;
提取该预选用户初始数据中的全部所述影响力特征的标签，所述影响力特征的标签包括：明星标签、平台认证标签、电商合作标签、创作类型标签和优质作品标签中的至少一种；Extracting all the labels of the influence characteristics in the initial data of the pre-selected user, wherein the labels of the influence characteristics include: at least one of: a celebrity label, a platform certification label, an e-commerce cooperation label, a creation type label, and a high-quality work label;
提取该预选用户初始数据中的粉丝特征，所述粉丝特征包括：该预选用户的每个粉丝的转发次数、评论次数和点赞次数；Extracting fan features from the initial data of the pre-selected user, wherein the fan features include: the number of reposts, comments, and likes of each fan of the pre-selected user;
提取该预选用户初始数据中的情感特征，所述情感特征包括：该预选用户收到的评论语句中的句子矩阵；Extracting sentiment features from the initial data of the pre-selected user, the sentiment features comprising: a sentence matrix in the comment sentence received by the pre-selected user;
提取该预选用户初始数据中的传播度特征，所述传播度特征包括：该预选用户全部作品的点赞数量、浏览数量、评论数量、被分享数量和粉丝数量。Extract the spreadability features in the initial data of the pre-selected user, wherein the spreadability features include: the number of likes, the number of views, the number of comments, the number of shares and the number of fans of all the works of the pre-selected user.
进一步地，所述利用所述关系特征，对全部所述预选用户执行聚类算法进行聚类，得到划分为不同类别的所述预选用户，包括：Furthermore, the utilizing the relationship features to perform clustering on all the pre-selected users using a clustering algorithm to obtain the pre-selected users divided into different categories includes:
利用全部所述预选用户的所述关系特征，对词向量模型进行训练，得到每个所述预选用户之间的距离；Using the relationship features of all the pre-selected users, training a word vector model to obtain the distance between each of the pre-selected users;
利用所述距离，对全部所述预选用户执行粗聚类算法进行粗聚类，得到质心个数并确定相应数目的质心；Utilizing the distance, executing a coarse clustering algorithm to perform coarse clustering on all the pre-selected users, obtaining the number of centroids and determining a corresponding number of centroids;
利用所述质心，对全部所述预选用户采取K类均值算法进行聚类，得到多个包含不同所述预选用户的团簇，根据所述团簇调整所述质心；Using the centroid, clustering all the pre-selected users using a K-means algorithm to obtain a plurality of clusters containing different pre-selected users, and adjusting the centroid according to the clusters;
循环执行K类均值算法进行聚类，至所述质心保持稳定；Circularly executing the K-means algorithm for clustering until the centroid remains stable;
将所述质心稳定的所述团簇作为所述类别，并将每个稳定的所述团簇中的所述预选用户划分为一类。The clusters with stable centroids are taken as the categories, and the pre-selected users in each stable cluster are divided into one category.
进一步地，所述采取极端梯度提升算法该候选用户的所述影响力特征进行加权，得到影响力得分，包括：Furthermore, the influence features of the candidate user are weighted by using the extreme gradient boosting algorithm to obtain an influence score, including:
对于每个所述候选用户，确定在该候选用户的所述影响力特征中所具备的所述标签；For each of the candidate users, determining the label included in the influence feature of the candidate user;
基于该候选用户具备的所述标签，采取极端梯度提升算法对该候选用户是否能够合法持有其余缺失的所述标签进行预测，得到预测结果；Based on the labels possessed by the candidate user, an extreme gradient boosting algorithm is used to predict whether the candidate user can legally possess the remaining missing labels, and a prediction result is obtained;
基于预测结果，利用该候选用户具备的所述标签和能够合法持有的其他所述标签计算该用户的影响力得分。Based on the prediction result, the influence score of the user is calculated using the tags possessed by the candidate user and other tags that can be legally held.
进一步地，所述利用所述粉丝特征计算粉丝重要度得分，包括：Further, the calculating the fan importance score by using the fan feature includes:
对于每个所述候选用户，利用该候选用户的每个所述粉丝的所述粉丝特征和该候选用户在获取周期内发布作品数量，计算该候选用户与该粉丝之间的信息量；For each candidate user, using the fan feature of each fan of the candidate user and the number of works published by the candidate user during the acquisition period, calculate the amount of information between the candidate user and the fan;
确定该粉丝与其关注的全部所述预选用户之间的信息量总量；Determine the total amount of information between the fan and all the pre-selected users followed by the fan;
利用关于该粉丝的所述信息量和所述信息量总量，确定该粉丝对于该候选用户的重视度；Determine the importance of the fan to the candidate user by using the amount of information about the fan and the total amount of information;
将全部所述候选用户与每个所述粉丝之间的重视度组成转移矩阵，并为所述转移矩阵赋予重要性向量，并基于预设的阻尼系数，构建关于所述重要性向量的迭代函数；The importance between all the candidate users and each of the fans is used to form a transfer matrix, and an importance vector is assigned to the transfer matrix, and an iterative function about the importance vector is constructed based on a preset damping coefficient;
响应于所述重要性向量在迭代中收敛至稳定，将该重要性向量确定为粉丝重要度得分。In response to the importance vector converging to stability during iteration, the importance vector is determined as a fan importance score.
进一步地，所述利用自回归语言模型、循环卷积神经网络和注意力机制构建情感分析模型，对所述情感特征进行概率分析，得到情感得分，包括：Furthermore, the sentiment analysis model is constructed by using an autoregressive language model, a recurrent convolutional neural network and an attention mechanism, and the sentiment features are subjected to probability analysis to obtain a sentiment score, including:
将所述情感特征输入所述子回归语言模型进行训练，得到文本动态特征；Inputting the sentiment feature into the sub-regression language model for training to obtain text dynamic features;
将所述文本动态特征输入所述循环卷积神经网络进行训练，得到深层的文本语义特征，其中，所述文本语义特征包括多个词向量；Inputting the text dynamic features into the recurrent convolutional neural network for training to obtain deep text semantic features, wherein the text semantic features include multiple word vectors;
对所述文本语义特征中的每个所述词向量采取所述注意力机制进行处理，得到每个所述词向量的语义权重；Using the attention mechanism to process each of the word vectors in the text semantic features to obtain a semantic weight of each of the word vectors;
对全部的所述语义权重和所述文本语义特征进行融合，得到所述情感得分。All the semantic weights and the text semantic features are fused to obtain the sentiment score.
进一步地，所述基于全部所述候选用户，利用所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分进行区间模糊处理，得到目标权重，包括：Furthermore, based on all the candidate users, interval fuzzy processing is performed using the communication characteristics, the influence score, the fan importance score and the sentiment score to obtain a target weight, including:
在全部所述候选用户中，将所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分均取最大值，并作为理想方案；Among all the candidate users, the maximum values of the communication feature, the influence score, the fan importance score and the sentiment score are taken as the ideal solution;
将所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分均取最小值，并作为临界方案；Taking the minimum value of the communication feature, the influence score, the fan importance score and the sentiment score as the critical solution;
对于每个所述候选用户，将该候选用户所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分的取值作为当前方案；For each candidate user, the values of the communication feature, the influence score, the fan importance score and the sentiment score of the candidate user are used as the current solution;
对于每个所述候选用户，利用所述理想方案、所述临界方案和所述当前方案构建投影模型；For each of the candidate users, construct a projection model using the ideal solution, the critical solution and the current solution;
基于最大熵原理，融合所述理想方案、所述临界方案和全部所述候选用户的所述当前方案，计算目标权重。Based on the maximum entropy principle, the ideal solution, the critical solution and the current solutions of all the candidate users are integrated to calculate the target weight.
基于同一发明构思，本申请还提供了一种自媒体用户选择装置，包括：Based on the same inventive concept, the present application also provides a self-media user selection device, including:
预处理模块，被配置为：在自媒体平台中周期性获取多个预选用户的初始数据，通过对所述初始数据进行预处理，得到每个所述预选用户的关系特征、影响力特征、粉丝特征、情感特征和传播度特征；The preprocessing module is configured to: periodically obtain initial data of multiple pre-selected users in the self-media platform, and obtain relationship characteristics, influence characteristics, fan characteristics, emotion characteristics and communication characteristics of each pre-selected user by pre-processing the initial data;
筛选模块，被配置为：利用所述关系特征，对全部所述预选用户执行聚类算法进行聚类，得到划分为不同类别的所述预选用户，在每个所述类别的所述预选用户中，按照预设比例确定多个候选用户；The screening module is configured to: utilize the relationship features to perform a clustering algorithm on all the pre-selected users to cluster them, obtain the pre-selected users divided into different categories, and determine a plurality of candidate users in the pre-selected users of each category according to a preset ratio;
多准则评价模块，被配置为：对全部所述候选用户中的每一个，采取极端梯度提升算法对该候选用户的所述影响力特征进行加权，得到影响力得分；利用所述粉丝特征计算粉丝重要度得分；利用自回归语言模型、循环卷积神经网络和注意力机制构建情感分析模型，对所述情感特征进行概率分析，得到情感得分；The multi-criteria evaluation module is configured to: for each of the candidate users, use the extreme gradient boosting algorithm to weight the influence feature of the candidate user to obtain an influence score; use the fan feature to calculate the fan importance score; use the autoregressive language model, recurrent convolutional neural network and attention mechanism to build a sentiment analysis model, perform probability analysis on the sentiment feature, and obtain a sentiment score;
排序模块，被配置为：基于全部所述候选用户，利用所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分进行区间模糊处理，得到目标权重，将每个所述候选用户的所述目标权重输入预置的多准则决策框架，得到该候选用户的综合值，并根据所述综合值确定目标用户。The sorting module is configured to: based on all the candidate users, perform interval fuzzy processing using the spreadability characteristics, the influence score, the fan importance score and the sentiment score to obtain a target weight, input the target weight of each candidate user into a preset multi-criteria decision framework to obtain a comprehensive value of the candidate user, and determine the target user based on the comprehensive value.
基于同一发明构思，本申请还提供了一种电子设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如上任意一项所述的自媒体用户选择方法。Based on the same inventive concept, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a self-media user selection method as described in any one of the above items is implemented.
基于同一发明构思，本申请还提供了一种非暂态计算机可读存储介质，其中，所述非暂态计算机可读存储介质存储计算机指令，所述计算机指令用于使所述计算机执行如上述自媒体用户选择方法。Based on the same inventive concept, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the self-media user selection method as described above.
从上面所述可以看出，本申请提供的自媒体用户选择方法及相关装置，基于对用户数据的整理，综合考虑了代表用户之间关系的关系特征、代表用户影响力的影响力特征、代表用户粉丝对其重视程度的粉丝特征，并综合了用户作品的传播度特征，来对用户进行多准则评价，使得对用户的评价结果准确，从而实现精准挑选影响力较大的自媒体目标用户。From the above, it can be seen that the self-media user selection method and related devices provided by the present application are based on the collation of user data, comprehensively consider the relationship characteristics representing the relationship between users, the influence characteristics representing the influence of users, and the fan characteristics representing the degree of attention paid by the fans of users to them, and comprehensively consider the dissemination characteristics of user works to perform multi-criteria evaluation on users, so that the evaluation results of users are accurate, thereby realizing the precise selection of self-media target users with greater influence.
附图说明BRIEF DESCRIPTION OF THE DRAWINGS
为了更清楚地说明本申请或相关技术中的技术方案，下面将对实施例或相关技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
图1为本申请实施例的自媒体用户选择方法的流程图；FIG1 is a flow chart of a method for selecting a self-media user according to an embodiment of the present application;
图2为本申请实施例的自媒体用户选择装置模块示意图；FIG2 is a schematic diagram of a module of a self-media user selection device according to an embodiment of the present application;
图3为本申请实施例的情感分析模型示意图；FIG3 is a schematic diagram of a sentiment analysis model according to an embodiment of the present application;
图4为本申请实施例的投影模型示意图；FIG4 is a schematic diagram of a projection model according to an embodiment of the present application;
图5为本申请实施例的电子设备结构示意图。FIG. 5 is a schematic diagram of the structure of an electronic device according to an embodiment of the present application.
具体实施方式Detailed ways
为使本申请的目的、技术方案和优点更加清楚明白，以下结合具体实施例，并参照附图，对本申请进一步详细说明。In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.
需要说明的是，除非另外定义，本申请的实施例使用的技术术语或者科学术语应当为本申请所属领域内具有一般技能的人士所理解的通常意义。本申请的实施例中使用的“第一”、“第二”以及类似的词语并不表示任何顺序、数量或者重要性，而只是用来区分不同的组成部分。“包括”或者“包含”等类似的词语意指出现该词前面的元件或者物件涵盖出现在该词后面列举的元件或者物件及其等同，而不排除其他元件或者物件。It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be understood by people with ordinary skills in the field to which the present application belongs. The "first", "second" and similar words used in the embodiments of the present application do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprising" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects.
如背景技术部分所述，相关的自媒体用户选择方法方法还难以满足商家及相关人员对目标用户进行选择的需要。As described in the background technology section, the related self-media user selection method is still difficult to meet the needs of merchants and related personnel to select target users.
申请人在实现本申请的过程中发现，相关的自媒体用户选择方法方法存在的主要问题在于：在自媒体平台中的各个用户，其所具备的特征多种多样，不同粉丝对于同一用户的粘性和重视程度不同，并且，申请人还发现，自媒体用户的影响力，传播能力可以从多维度进行评价。In the process of implementing this application, the applicant found that the main problem with the relevant self-media user selection method is that each user in the self-media platform has a variety of characteristics, and different fans have different stickiness and attention to the same user. In addition, the applicant also found that the influence and communication ability of self-media users can be evaluated from multiple dimensions.
可以理解，该方法可以通过任何具有计算、处理能力的装置、设备、平台、设备集群来执行。It can be understood that the method can be executed by any device, equipment, platform, or device cluster with computing and processing capabilities.
以下，通过具体的实施例，来详细说明本申请的技术方法。The technical method of the present application is described in detail below through specific embodiments.
参考图1，本申请一个实施例的自媒体用户选择方法，包括以下步骤：Referring to FIG1 , a method for selecting a self-media user according to an embodiment of the present application includes the following steps:
步骤S101、在自媒体平台中周期性获取多个预选用户的初始数据，通过对所述初始数据进行预处理，得到每个所述预选用户的关系特征、影响力特征、粉丝特征、情感特征和传播度特征。Step S101: periodically obtain initial data of multiple pre-selected users in the self-media platform, and obtain relationship characteristics, influence characteristics, fan characteristics, emotion characteristics and communication characteristics of each pre-selected user by pre-processing the initial data.
在本申请的实施例中，在确定的自媒体平台中，根据预设的周期，周期性地获取多个预选用户在该自媒体平台中的数据作为初始数据，并进行预处理。In an embodiment of the present application, in a determined self-media platform, data of a plurality of pre-selected users in the self-media platform is periodically obtained as initial data according to a preset period, and pre-processed.
具体地，对于每个预选用户，所获取的初始数据可以包括：Specifically, for each pre-selected user, the initial data obtained may include:
该预选用户在该自媒体平台中的身份识别标签、昵称；The identity identification label and nickname of the pre-selected user in the self-media platform;
与该预选用户互相关注的其他预选用户，并将其作为该预选用户的关系特征；Other pre-selected users who mutually follow the pre-selected user are used as relationship features of the pre-selected user;
该预选用户拥有的部分或全部的明星标签、平台认证标签、电商合作标签、创作类型标签和优质作品标签，并将其作为该预选用户的影响力特征；Some or all of the celebrity labels, platform certification labels, e-commerce cooperation labels, creative type labels, and high-quality work labels that the pre-selected user possesses, and use them as the influence characteristics of the pre-selected user;
该预选用户的每个粉丝的转发次数、评论次数和点赞次数，并将其作为该预选用户的粉丝特征；The number of reposts, comments and likes of each fan of the pre-selected user is used as the fan feature of the pre-selected user;
该预选用户收到的评论语句，并将其作为该预选用户的情感特征；The comment sentence received by the pre-selected user is used as the sentiment feature of the pre-selected user;
该预选用户全部作品中的点赞数量、浏览数量、评论数量、被分享数量和粉丝数量，并将其作为该预选用户的传播度特征。The number of likes, views, comments, shares and fans of all works of the pre-selected user are used as the communication characteristics of the pre-selected user.
需要说明的是，在本实施例中，最终所要选择的目标用户为影响力较大的自媒体创作者，但由于自媒体平台中的非创作者用户与创作者用户之间界限并不固定，并且不清晰，因此将所有用户作为基础进行筛选。It should be noted that, in this embodiment, the target users to be ultimately selected are self-media creators with greater influence. However, since the boundary between non-creator users and creator users in the self-media platform is not fixed and clear, all users are used as the basis for screening.
步骤S102、利用所述关系特征，对全部所述预选用户执行聚类算法进行聚类，得到划分为不同类别的所述预选用户，在每个所述类别的所述预选用户中，按照预设比例确定多个候选用户。Step S102: using the relationship features, executing a clustering algorithm to cluster all the pre-selected users to obtain the pre-selected users divided into different categories, and determining a plurality of candidate users in the pre-selected users of each category according to a preset ratio.
在本申请的实施例中，基于上述的预选用户，综合利用多种聚类算法筛选出候选用户，以缩小选择目标用户的范围。In the embodiment of the present application, based on the above-mentioned pre-selected users, multiple clustering algorithms are comprehensively utilized to screen out candidate users, so as to narrow down the scope of selecting target users.
具体地，将相似程度高的预选用户聚为一类，并在每类相似的用户中选择一定比例的预选用户作为候选用户。Specifically, pre-selected users with high similarity are grouped into one category, and a certain proportion of pre-selected users are selected from each category of similar users as candidate users.
其中，相似的预选用户之间往往存在社交关系，通过分析社交关系，为了将相似程度高的预选用户聚类在一起，因此需要计算预选用户之间的距离。There are often social relationships between similar pre-selected users. By analyzing the social relationships, in order to cluster the pre-selected users with high similarity, the distance between the pre-selected users needs to be calculated.
在本实施例中，将预选用户之间存在相互关注的社交关系，定义为他们之间是熟悉的，并有聚类特性的。In this embodiment, the social relationship of mutual attention between the pre-selected users is defined as being familiar with each other and having clustering characteristics.
因此，在本实施例中，利用预选用户之间的相互关注关系，以分词的处理方式，封装成数据集。Therefore, in this embodiment, the mutual attention relationship between pre-selected users is utilized and encapsulated into a data set in a word segmentation processing manner.
具体地，对于每个预选用户，将关系特征作为输入文本，输入至Word2Vec(词向量)模型中，并对该Word2Vec模型进行训练，得到关于该预选用户的多维参数，并将该多维参数作为多维的参数向量，利用该多维参数向量计算该预选用户与其他预选用户之间的距离，在本实施例中，所述距离可以是欧氏距离。Specifically, for each pre-selected user, the relationship features are taken as input text and input into a Word2Vec (word vector) model, and the Word2Vec model is trained to obtain multi-dimensional parameters about the pre-selected user, and the multi-dimensional parameters are taken as a multi-dimensional parameter vector, and the distance between the pre-selected user and other pre-selected users is calculated using the multi-dimensional parameter vector. In this embodiment, the distance may be a Euclidean distance.
进一步地，利用Canopy(粗聚类)算法，对预选用户进行粗聚类。Furthermore, the Canopy (coarse clustering) algorithm is used to perform coarse clustering on the pre-selected users.
具体地，对于该预选用户预设第一距离阈值，将在该第一距离阈值之内的其他预选用户和该预选用户均作为数据点输入至Canopy算法。Specifically, a first distance threshold is preset for the pre-selected user, and the pre-selected user and other pre-selected users within the first distance threshold are input into the Canopy algorithm as data points.
进一步地，得到Canopy算法输出的团簇，其中，团簇表征了聚类，团簇的个数为聚类个数，也即质心的个数，并且团簇之间可以重叠。Furthermore, the clusters output by the Canopy algorithm are obtained, where the clusters represent clusters, the number of clusters is the number of clusters, that is, the number of centroids, and the clusters can overlap.
基于上述得到的团簇，对预选用户执行K-Means(K类均值)算法，以得到更加细致的聚类。Based on the clusters obtained above, the K-Means algorithm is executed on the pre-selected users to obtain more detailed clustering.
具体地，将上述Canopy算法得到的质心个数作为K-Means算法的聚类质心个数，并按照上述预选用户之间的距离，将其分配到各个K-Means算法中的团簇中，在全部预选用户分配完成后，重新确定每个聚类的质心，并根据重新确定的质心，再次对预选用户执行K-Means算法，以重新进行聚类，在多次的聚类过程中，每个聚类的质心和聚类中所包括的预选用户将被不断调整。Specifically, the number of centroids obtained by the above Canopy algorithm is used as the number of cluster centroids of the K-Means algorithm, and the pre-selected users are allocated to clusters in the K-Means algorithm according to the distances between them. After all the pre-selected users are allocated, the centroid of each cluster is re-determined, and the K-Means algorithm is executed again on the pre-selected users based on the re-determined centroid to re-cluster them. In multiple clustering processes, the centroid of each cluster and the pre-selected users included in the cluster will be continuously adjusted.
进一步地，当聚类质心不再发生变化，保持稳定，即代表了聚类也即团簇中的预选用户不在发生变化，保持稳定，可以将此时的聚类结果作为本步骤中最终的聚类结果。Furthermore, when the cluster centroid no longer changes and remains stable, it means that the cluster, ie, the pre-selected users in the cluster, no longer change and remain stable. The clustering result at this time can be used as the final clustering result in this step.
在本实施例中，根据上述确定的聚类结果，在每个团簇中按照预定的比例和预定的策略确定候选用户，例如，将影响力特征最高的前20％的预选用户作为候选用户，或者将其他特征最高的前30％或其他比例的预选用户确定为候选用户。In this embodiment, based on the clustering results determined above, candidate users are determined in each cluster according to a predetermined ratio and a predetermined strategy. For example, the top 20% of pre-selected users with the highest influence characteristics are determined as candidate users, or the top 30% or other ratios of pre-selected users with the highest other characteristics are determined as candidate users.
步骤S103、对全部所述候选用户中的每一个，采取极端梯度提升算法对该候选用户的所述影响力特征进行加权，得到影响力得分；利用所述粉丝特征计算粉丝重要度得分；利用自回归语言模型、循环卷积神经网络和注意力机制构建情感分析模型，对所述情感特征进行概率分析，得到情感得分。Step S103: For each of the candidate users, use the extreme gradient boosting algorithm to weight the influence characteristics of the candidate user to obtain an influence score; use the fan characteristics to calculate the fan importance score; use the autoregressive language model, recurrent convolutional neural network and attention mechanism to build a sentiment analysis model, perform probability analysis on the sentiment characteristics, and obtain a sentiment score.
在本申请的实施例中，采取了包括影响力得分、粉丝重要度得分和情感得分在内的多准则评价的方式，对目标用户进行选择，对于每个得分准则的计算，在执行上没有先后顺序，在本实施例中的描述顺序仅仅作为示例。In the embodiments of the present application, a multi-criteria evaluation method including influence score, fan importance score and sentiment score is adopted to select target users. There is no order of execution for the calculation of each scoring criterion, and the description order in this embodiment is only for example.
在本实施例中，对于每个候选用户，在影响力特征：明星标签、平台认证标签、电商合作标签、创作类型标签和优质作品标签中，确定该候选用户所具备的标签。In this embodiment, for each candidate user, the labels possessed by the candidate user are determined among the influence features: celebrity label, platform certification label, e-commerce cooperation label, creation type label, and high-quality work label.
由于在实际情况中，往往存在某个候选用户只具备其中的一部分标签，但具备持有其他缺失标签的资格，由于自媒体平台认证迟缓，或者其他原因，导致该候选用户没有持有部分标签，因此首先需要对候选用户判定是否有资格持有其他缺失的标签。In actual situations, there is often a candidate user who only has some of the tags, but is qualified to hold other missing tags. Due to slow authentication of self-media platforms or other reasons, the candidate user does not hold some tags. Therefore, it is first necessary to determine whether the candidate user is qualified to hold other missing tags.
具体地，将候选用户所具备的标签输入至XGBoost(极端梯度提升算法)的回归树模型中进行训练。Specifically, the labels possessed by the candidate users are input into the regression tree model of XGBoost (extreme gradient boosting algorithm) for training.
其中，在本实施例中的XGBoost有强大泛化能力，并且具备足够准确的树群。Among them, XGBoost in this embodiment has strong generalization ability and has sufficiently accurate tree groups.
进一步地，XGBoost回归树模型通过对候选用户所持有的标签预测该候选用户是否具备持有其他缺失标签的资格。Furthermore, the XGBoost regression tree model predicts whether the candidate user is qualified to hold other missing tags based on the tags held by the candidate user.
若该候选用户具备持有某个或某几个缺失的标签的资格，则将该标签与其所持有的标签共同进行加权融合，并将融合结果作为影响力得分。If the candidate user is qualified to hold one or several missing tags, the tag is weightedly fused with the tags held by the candidate user, and the fusion result is used as the influence score.
若该候选用户不具备持有其他我缺失的标签的资格，则只将其所持有的标进行加权融合，并将融合结果作为影响力得分。If the candidate user does not have the qualifications to hold other missing tags, only the tags he holds will be weighted and fused, and the fusion result will be used as the influence score.
在本实施例中，自媒体平台具有一定的社交属性，可以根据候选用户间的关注和被关注关系构建复杂网络模型，为了反应粉丝对短视频创作者的重视程度，引入了PageRank(网页排名)思想，通过不断迭代得到粉丝影重要度得分。In this embodiment, the self-media platform has certain social attributes, and a complex network model can be constructed based on the follow-up and followed relationships between candidate users. In order to reflect the degree of attention fans pay to short video creators, the PageRank (web page ranking) idea is introduced, and the fan influence importance score is obtained through continuous iteration.
具体地，如步骤S101所述，每个粉丝的转发次数、评论次数和点赞次数，并将其作为该候选用户的粉丝特征，因此，对于每个候选用户，确定候选用户的粉丝特征中各项数值。Specifically, as described in step S101, the number of reposts, comments and likes of each fan is used as the fan feature of the candidate user. Therefore, for each candidate user, the values of each item in the fan feature of the candidate user are determined.
进一步地，对于每个候选用户，利用粉丝特征，采取如下公式确定该候选用户与每个分子之间的信息量：Furthermore, for each candidate user, the following formula is used to determine the amount of information between the candidate user and each molecule using the fan feature:
其中，i表示候选用户，j表示粉丝，Rt(i,j)表示在获取周期内粉丝j对候选用户i的作品的转发次数、评论次数和点赞次数的综合，S(i)表示在获取周期内候选用户i所发布的作品的数量。Where i represents the candidate user, j represents the fan, Rt(i,j) represents the total number of reposts, comments, and likes of the works of candidate user i by fan j during the acquisition period, and S(i) represents the number of works published by candidate user i during the acquisition period.
进一步地，利用如下公式确定上述的粉丝j对于候选用户i的重视度：Furthermore, the following formula is used to determine the importance of the above-mentioned fan j to the candidate user i:
其中，p表示粉丝i所关注的其他某个用户，U则为该粉丝i所关注的其他用户的全部集合；也即，该公式的分子表征了该候选用户与该分子之间的传递的信息量，而分母表征了该粉丝与全部用户之间的信息量总量。Among them, p represents another user followed by fan i, and U is the entire set of other users followed by fan i; that is, the numerator of the formula represents the amount of information transmitted between the candidate user and the numerator, while the denominator represents the total amount of information between the fan and all users.
进一步地，利用上述算法，计算候选用户i的每个粉丝对于该候选用户的重视度，并进一步计算每个候选用户与其每个粉丝之间的重视度。Furthermore, the algorithm is used to calculate the importance of each fan of candidate user i to the candidate user, and further calculate the importance between each candidate user and each of its fans.
进一步地，将计算出的每个重视度作为一个元素，组成转移矩阵M，并设计重要性向量，利用如下所示的迭代函数对重要性向量进行迭代：Furthermore, each calculated importance is used as an element to form a transfer matrix M, and an importance vector is designed. The importance vector is iterated using the iterative function shown below:
其中，V表示前一次迭代中的重要性向量，并在初次迭代时，对重要性向量进行初始化作为公式中的V，V’表示本次迭代所计算出的重要性向量，e为单位向量，N为网络中节点的数量，d为预先设置的阻尼系数。Wherein, V represents the importance vector in the previous iteration, and in the first iteration, the importance vector is initialized as V in the formula, V' represents the importance vector calculated in this iteration, e is the unit vector, N is the number of nodes in the network, and d is the pre-set damping coefficient.
进一步地，经过多次迭代，重要性向量的变化由于阻尼系数的存在将逐渐收敛，当重要性向量趋于稳定，并不再变化时，将此时的重要性向量确定为粉丝重要度得分。Furthermore, after multiple iterations, the change of the importance vector will gradually converge due to the existence of the damping coefficient. When the importance vector tends to be stable and no longer changes, the importance vector at this time is determined as the fan importance score.
在本实施例中，为了确定情感得分，可以利用XLNet 302(自回归语言模型)、Bi-SRU 303(循环神经网络)、注意力机制和Softmax函数(也即分类器)构建情感分析模型。In this embodiment, in order to determine the sentiment score, a sentiment analysis model may be constructed using XLNet 302 (autoregressive language model), Bi-SRU 303 (recurrent neural network), attention mechanism and Softmax function (ie, classifier).
具体地，如图3所示，对于每个候选用户，将其全部作品所收到的评论语句，也即情感特征作为句子矩阵，在图3中表示为X1，X2……Xn，由输出层301输入至XLNet 302进行训练，以充分挖掘长文本间的信息，并保存文本的文本动态特征T。Specifically, as shown in FIG3 , for each candidate user, the comment sentences received for all his works, that is, the sentiment features, are taken as a sentence matrix, represented as X1 , X2 , …Xn in FIG3 , and input into XLNet 302 through the output layer 301 for training, so as to fully mine the information between long texts and save the text dynamic features T of the texts.
进一步地，将得到的文本动态特征输入至Bi-SRU 303，使用Bi-SRU 303对文本动态特征进行训练，在缩短训练时间的基础上进一步提取深层的文本语义特征，其中，每个文本语义特征均可以是一个词向量。Furthermore, the obtained text dynamic features are input into Bi-SRU 303, and Bi-SRU 303 is used to train the text dynamic features, so as to further extract deep text semantic features on the basis of shortening the training time, wherein each text semantic feature can be a word vector.
进一步地，将深层的文本语义特征送入注意力机制模型304中并使用Softmax函数进行处理，以过滤无关信息，解决信息过载的问题。Furthermore, the deep text semantic features are fed into the attention mechanism model 304 and processed using the Softmax function to filter out irrelevant information and solve the problem of information overload.
进一步地，通过注意力机制可以得到上述词向量的权重信息h，具体地，可以是一个矩阵，并称为语义权重矩阵，将语义权重矩阵中的每语义权重h作为对应所述词向量a的权重。Furthermore, the weight information h of the above-mentioned word vector can be obtained through the attention mechanism. Specifically, it can be a matrix, and is called a semantic weight matrix. Each semantic weight h in the semantic weight matrix is used as the weight corresponding to the word vector a.
进一步地，将上述得到的语义权重输入至输出层305，以将该语义权重与深层的文本语义特征，也即词向量对应相乘并进行求和运算得到每条文本数据在情感类别中的条件概率，并将其作为该候选用户的情感得分。其中，情感的类别可以包括：例如，喜爱和讨厌两类。Furthermore, the semantic weight obtained above is input to the output layer 305, so as to multiply the semantic weight with the deep text semantic feature, that is, the word vector, and perform a sum operation to obtain the conditional probability of each text data in the sentiment category, and use it as the sentiment score of the candidate user. Among them, the sentiment category can include: for example, like and hate.
步骤S104、基于全部所述候选用户，利用所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分进行区间模糊处理，得到目标权重，将每个所述候选用户的所述目标权重输入预置的多准则决策框架，得到该候选用户的综合值，并根据所述综合值确定目标用户。Step S104: Based on all the candidate users, interval fuzzy processing is performed using the communication characteristics, the influence score, the fan importance score and the sentiment score to obtain a target weight, and the target weight of each candidate user is input into a preset multi-criteria decision framework to obtain a comprehensive value of the candidate user, and the target user is determined based on the comprehensive value.
在本申请的实施例中，基于上述方法，获取全部候选用户的传播度特征，计算全部候选用户的影响力得分、粉丝重要度得分和情感得分，并将传播度特征、影响力得分、粉丝重要度得分和情感得分同时作为多准则评价的多项指标。In an embodiment of the present application, based on the above method, the communication characteristics of all candidate users are obtained, the influence scores, fan importance scores and sentiment scores of all candidate users are calculated, and the communication characteristics, influence scores, fan importance scores and sentiment scores are used as multiple indicators for multi-criteria evaluation.
进一步地，构建如图4所示的模糊区间的投影模型。Furthermore, a projection model of the fuzzy interval as shown in FIG4 is constructed.
具体地，在上述计算出的全部候选用户中，选择传播度特征的最大值、影响力得分的最大值、粉丝重要度得分的最大值和情感得分均的最大值作为理想方案，并在本实施例中表示为X+。Specifically, among all the candidate users calculated above, the maximum value of the propagation feature, the maximum value of the influence score, the maximum value of the fan importance score and the maximum value of the emotion score are selected as the ideal solution, and are represented as X+ in this embodiment.
进一步地，在上述计算出的全部候选用户中，选择传播度特征的最小值、影响力得分的小值、粉丝重要度得分的最小值和情感得分均的最小值作为临界方案，并在本实施例中表示为X-。Furthermore, among all the candidate users calculated above, the minimum value of the propagation feature, the minimum value of the influence score, the minimum value of the fan importance score and the minimum value of the sentiment score are selected as the critical solution, and are represented as X- in this embodiment.
进一步地，将每个候选用户的传播度特征取值、影响力得分取值、粉丝重要度得分取值和情感得分取值作为当前方案，并在本实施例中表示为Xi。Furthermore, the propagation characteristic value, influence score value, fan importance score value and sentiment score value of each candidate user are taken as the current scheme and are represented asXi in this embodiment.
进一步地，根据图4所示，将理想方案、临界方案和当前方案构成向量X-Xi、向量X+Xi和向量X-X+，其中，向量x+xi在X-X+上的投影表示为向量X-Xi在X-X+上的投影/>Further, as shown in FIG4 , the ideal solution, the critical solution and the current solution constitute vector X-Xi , vector X+Xi and vector XX+ , where the projection of vector x+xi on X- X+ is expressed as Projection of vector X-Xi on X- X+ />
进一步地，利用如下所示的第一投影公式计算Further, the first projection formula is used to calculate
利用如下所示的第二投影公式计算Use the second projection formula as shown below to calculate
其中，在第一投影公式和第二投影公式中，C均表示协方差，E均表示期望。In the first projection formula and the second projection formula, C represents covariance, and E represents expectation.
进一步地，为使当前方案在理想方案上的投影最大，在临界方案上的投影最小，在本实施例中引入最大熵原理，构建了如下的目标函数：Furthermore, in order to maximize the projection of the current solution on the ideal solution and minimize the projection on the critical solution, the maximum entropy principle is introduced in this embodiment to construct the following objective function:
其中，为/>的第一权重集合，/>为/>的第二权重集合，n为全部候选用户，m为多准则评价中的全部指标。in, For/> The first weight set of For/> The second weight set is , n is all candidate users, and m is all indicators in the multi-criteria evaluation.
进一步地，当上述目标函数的结果达到最大时，确定此时的第一权重集合和第二权重集合，并共同作为目标权重的两个权重分量。Furthermore, when the result of the above objective function reaches the maximum, the first weight set and the second weight set at this time are determined and used together as two weight components of the target weight.
进一步地，对于每个候选用户，将得到的上述指标进行加成求和，并通上述目标权重一起输入至VIKOR(多准则决策框架)中进行计算，得到该候选用户的综合值，并在全部候选用户中，根据综合值进行排序并优选所需个数的候选用户作为目标用户。Furthermore, for each candidate user, the above indicators are added and summed, and input into VIKOR (multi-criteria decision framework) together with the above target weight for calculation to obtain the comprehensive value of the candidate user, and among all the candidate users, they are sorted according to the comprehensive value and the required number of candidate users are selected as target users.
可见，本申请的实施例的自媒体用户选择方法，基于对用户数据的整理，综合考虑了代表用户之间关系的关系特征、代表用户影响力的影响力特征、代表用户粉丝对其重视程度的粉丝特征，并综合了用户作品的传播度特征，来对用户进行多准则评价，使得对用户的评价结果准确，从而实现精准挑选影响力较大的自媒体目标用户。It can be seen that the self-media user selection method of the embodiment of the present application, based on the collation of user data, comprehensively considers the relationship characteristics representing the relationship between users, the influence characteristics representing the influence of users, the fan characteristics representing the degree of attention paid by the users' fans to them, and comprehensively considers the dissemination characteristics of the users' works to perform multi-criteria evaluation on users, so that the evaluation results of users are accurate, thereby realizing the precise selection of self-media target users with greater influence.
需要说明的是，本申请的实施例的方法可以由单个设备执行，例如一台计算机或服务器等。本实施例的方法也可以应用于分布式场景下，由多台设备相互配合来完成。在这种分布式场景的情况下，这多台设备中的一台设备可以只执行本申请的实施例的方法中的某一个或多个步骤，这多台设备相互之间会进行交互以完成所述的方法。It should be noted that the method of the embodiment of the present application can be performed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the described method.
需要说明的是，上述对本申请的一些实施例进行了描述。其它实施例在所附权利要求书的范围内。在一些情况下，在权利要求书中记载的动作或步骤可以按照不同于上述实施例中的顺序来执行并且仍然可以实现期望的结果。另外，在附图中描绘的过程不一定要求示出的特定顺序或者连续顺序才能实现期望的结果。在某些实施方式中，多任务处理和并行处理也是可以的或者可能是有利的。It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
基于同一发明构思，与上述任意实施例方法相对应的，本申请的实施例还提供了一种自媒体用户选择装置。Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, an embodiment of the present application further provides a self-media user selection device.
参考图2，所述自媒体用户选择装置，包括：预处理模块201、筛选模块202、多准则评价模块203和排序模块204。2 , the self-media user selection device includes: a pre-processing module 201 , a screening module 202 , a multi-criteria evaluation module 203 and a sorting module 204 .
其中，预处理模块201，被配置为：在自媒体平台中周期性获取多个预选用户的初始数据，通过对所述初始数据进行预处理，得到每个所述预选用户的关系特征、影响力特征、粉丝特征、情感特征和传播度特征；The preprocessing module 201 is configured to: periodically obtain initial data of multiple pre-selected users in the self-media platform, and obtain the relationship characteristics, influence characteristics, fan characteristics, emotion characteristics and communication characteristics of each pre-selected user by pre-processing the initial data;
筛选模块202，被配置为：利用所述关系特征，对全部所述预选用户执行聚类算法进行聚类，得到划分为不同类别的所述预选用户，在每个所述类别的所述预选用户中，按照预设比例确定多个候选用户；The screening module 202 is configured to: perform a clustering algorithm on all the pre-selected users to cluster them using the relationship features, obtain the pre-selected users divided into different categories, and determine a plurality of candidate users in each category of the pre-selected users according to a preset ratio;
多准则评价模块203，被配置为：对全部所述候选用户中的每一个，采取极端梯度提升算法对该候选用户的所述影响力特征进行加权，得到影响力得分；利用所述粉丝特征计算粉丝重要度得分；利用自回归语言模型、循环卷积神经网络和注意力机制构建情感分析模型，对所述情感特征进行概率分析，得到情感得分；The multi-criteria evaluation module 203 is configured to: for each of the candidate users, use the extreme gradient boosting algorithm to weight the influence feature of the candidate user to obtain an influence score; use the fan feature to calculate the fan importance score; use the autoregressive language model, recurrent convolutional neural network and attention mechanism to build a sentiment analysis model, perform probability analysis on the sentiment feature, and obtain a sentiment score;
排序模块204，被配置为：基于全部所述候选用户，利用所述传播度特征、所述影响力得分、所述粉丝重要度得分和所述情感得分进行区间模糊处理，得到目标权重，将每个所述候选用户的所述目标权重输入预置的多准则决策框架，得到该候选用户的综合值，并根据所述综合值确定目标用户。The sorting module 204 is configured to: based on all the candidate users, perform interval fuzzy processing using the spreadability characteristics, the influence score, the fan importance score and the sentiment score to obtain a target weight, input the target weight of each candidate user into a preset multi-criteria decision framework to obtain a comprehensive value of the candidate user, and determine the target user based on the comprehensive value.
为了描述的方便，描述以上装置时以功能分为各种模块分别描述。当然，在实施本申请的实施例时可以把各模块的功能在同一个或多个软件和/或硬件中实现。For the convenience of description, the above devices are described in terms of functions and are divided into various modules. Of course, when implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and/or hardware.
上述实施例的装置用于实现前述任一实施例中相应的自媒体用户选择方法，并且具有相应的方法实施例的有益效果，在此不再赘述。The device of the above embodiment is used to implement the corresponding self-media user selection method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.
基于同一发明构思，与上述任意实施例方法相对应的，本申请的实施例还提供了一种电子设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如上任意一实施例所述的自媒体用户选择方法。Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, an embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the self-media user selection method as described in any of the above embodiments is implemented.
图5示出了本实施例所提供的一种更为具体的电子设备硬件结构示意图，该设备可以包括：处理器1010、存储器1020、输入/输出接口1030、通信接口1040和总线1050。其中处理器1010、存储器1020、输入/输出接口1030和通信接口1040通过总线1050实现彼此之间在设备内部的通信连接。5 shows a more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment, and the device may include: a processor 1010, a memory 1020, an input/output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input/output interface 1030, and the communication interface 1040 are connected to each other in communication within the device through the bus 1050.
处理器1010可以采用通用的CPU(Central Processing Unit，中央处理器)、微处理器、应用专用集成电路(Application Specific Integrated Circuit，ASIC)、或者一个或多个集成电路等方式实现，用于执行相关程序，以实现本申请实施例所提供的技术方案。The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.
存储器1020可以采用ROM(Read Only Memory，只读存储器)、RAM(Random AccessMemory，随机存取存储器)、静态存储设备，动态存储设备等形式实现。存储器1020可以存储操作系统和其他应用程序，在通过软件或者固件来实现本申请实施例所提供的技术方案时，相关的程序代码保存在存储器1020中，并由处理器1010来调用执行。The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solution provided in the embodiment of the present application is implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.
输入/输出接口1030用于连接输入/输出模块，以实现信息输入及输出。输入输出/模块可以作为组件配置在设备中(图中未示出)，也可以外接于设备以提供相应功能。其中输入设备可以包括键盘、鼠标、触摸屏、麦克风、各类传感器等，输出设备可以包括显示器、扬声器、振动器、指示灯等。The input/output interface 1030 is used to connect the input/output module to realize information input and output. The input/output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.
通信接口1040用于连接通信模块(图中未示出)，以实现本设备与其他设备的通信交互。其中通信模块可以通过有线方式(例如USB、网线等)实现通信，也可以通过无线方式(例如移动网络、WIFI、蓝牙等)实现通信。The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).
总线1050包括一通路，在设备的各个组件(例如处理器1010、存储器1020、输入/输出接口1030和通信接口1040)之间传输信息。The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input/output interface 1030, and the communication interface 1040).
需要说明的是，尽管上述设备仅示出了处理器1010、存储器1020、输入/输出接口1030、通信接口1040以及总线1050，但是在具体实施过程中，该设备还可以包括实现正常运行所必需的其他组件。此外，本领域的技术人员可以理解的是，上述设备中也可以仅包含实现本申请实施例方案所必需的组件，而不必包含图中所示的全部组件。It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input/output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiment of the present application, and does not necessarily include all the components shown in the figure.
上述实施例的装置用于实现前述任一实施例中相应的自媒体用户选择方法，并且具有相应的方法实施例的有益效果，在此不再赘述。The device of the above embodiment is used to implement the corresponding self-media user selection method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.
基于同一发明构思，与上述任意实施例方法相对应的，本申请还提供了一种非暂态计算机可读存储介质，所述非暂态计算机可读存储介质存储计算机指令，所述计算机指令用于使所述计算机执行如上任一实施例所述的自媒体用户选择方法。Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the self-media user selection method described in any of the above embodiments.
本实施例的计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.
上述实施例的存储介质存储的计算机指令用于使所述计算机执行如上任一实施例所述的自媒体用户选择方法，并且具有相应的方法实施例的有益效果，在此不再赘述。The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the self-media user selection method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.
所属领域的普通技术人员应当理解：以上任何实施例的讨论仅为示例性的，并非旨在暗示本申请的范围(包括权利要求)被限于这些例子；在本申请的思路下，以上实施例或者不同实施例中的技术特征之间也可以进行组合，步骤可以以任意顺序实现，并存在如上所述的本申请的实施例的不同方面的许多其它变化，为了简明它们没有在细节中提供。Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.
另外，为简化说明和讨论，并且为了不会使本申请的实施例难以理解，在所提供的附图中可以示出或可以不示出与集成电路(IC)芯片和其它部件的公知的电源/接地连接。此外，可以以框图的形式示出装置，以便避免使本申请的实施例难以理解，并且这也考虑了以下事实，即关于这些框图装置的实施方式的细节是高度取决于将要实施本申请的实施例的平台的(即，这些细节应当完全处于本领域技术人员的理解范围内)。在阐述了具体细节(例如，电路)以描述本申请的示例性实施例的情况下，对本领域技术人员来说显而易见的是，可以在没有这些具体细节的情况下或者这些具体细节有变化的情况下实施本申请的实施例。因此，这些描述应被认为是说明性的而不是限制性的。In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power supply/ground connection with the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present application will be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). In the case of elaborating specific details (e.g., circuits) to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered illustrative rather than restrictive.
尽管已经结合了本申请的具体实施例对本申请进行了描述，但是根据前面的描述，这些实施例的很多替换、修改和变型对本领域普通技术人员来说将是显而易见的。例如，其它存储器架构(例如，动态RAM(DRAM))可以使用所讨论的实施例。Although the present application has been described in conjunction with specific embodiments of the present application, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.
本申请的实施例旨在涵盖落入所附权利要求的宽泛范围之内的所有这样的替换、修改和变型。因此，凡在本申请的实施例的精神和原则之内，所做的任何省略、修改、等同替换、改进等，均应包含在本申请的保护范围之内。The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the protection scope of the present application.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111641822.4ACN114580427B (en) | 2021-12-29 | 2021-12-29 | Self-media user selection method and related equipment |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111641822.4ACN114580427B (en) | 2021-12-29 | 2021-12-29 | Self-media user selection method and related equipment |
| Publication Number | Publication Date |
|---|---|
| CN114580427A CN114580427A (en) | 2022-06-03 |
| CN114580427Btrue CN114580427B (en) | 2024-06-14 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111641822.4AActiveCN114580427B (en) | 2021-12-29 | 2021-12-29 | Self-media user selection method and related equipment |
| Country | Link |
|---|---|
| CN (1) | CN114580427B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020116A (en)* | 2012-11-13 | 2013-04-03 | 中国科学院自动化研究所 | Method for automatically screening influential users on social media networks |
| CN107341270A (en)* | 2017-07-28 | 2017-11-10 | 东北大学 | Towards the user feeling influence power analysis method of social platform |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107862617A (en)* | 2017-10-20 | 2018-03-30 | 江苏大学 | A kind of microblogging community division method based on user's comprehensive similarity |
| CN108460499B (en)* | 2018-04-02 | 2022-03-08 | 福州大学 | Microblog user influence ranking method integrating user time information |
| CN109299420A (en)* | 2018-09-18 | 2019-02-01 | 精硕科技（北京）股份有限公司 | Social media account processing method, device, equipment and readable storage medium storing program for executing |
| CN110442857B (en)* | 2019-06-18 | 2024-05-10 | 平安科技（深圳）有限公司 | Emotion intelligent judging method and device and computer readable storage medium |
| CN112182152B (en)* | 2020-09-24 | 2021-09-03 | 东北大学 | Sina microblog user emotion influence analysis method based on deep learning |
| CN112487264B (en)* | 2020-11-27 | 2024-05-10 | 上海蜜度信息技术有限公司 | Method, system and device for mining key users based on social media content |
| CN113434783A (en)* | 2021-07-02 | 2021-09-24 | 北京中奥淘数据科技有限公司 | Method and device for calculating influence of network user and electronic equipment |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020116A (en)* | 2012-11-13 | 2013-04-03 | 中国科学院自动化研究所 | Method for automatically screening influential users on social media networks |
| CN107341270A (en)* | 2017-07-28 | 2017-11-10 | 东北大学 | Towards the user feeling influence power analysis method of social platform |
| Publication number | Publication date |
|---|---|
| CN114580427A (en) | 2022-06-03 |
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | EDMF: Efficient deep matrix factorization with review feature learning for industrial recommender system | |
| Joung et al. | Approach for importance–performance analysis of product attributes from online reviews | |
| JP6753115B2 (en) | Content management device, content management method and program | |
| Yang et al. | DOMFN: A divergence-orientated multi-modal fusion network for resume assessment | |
| Velmurugan et al. | Developing a fidelity evaluation approach for interpretable machine learning | |
| KR102410715B1 (en) | Apparatus and method for analyzing sentiment of text data based on machine learning | |
| Rathee et al. | Sentiment analysis using machine learning techniques on Python | |
| JP6004015B2 (en) | Learning method, information processing apparatus, and learning program | |
| CN112632226A (en) | Semantic search method and device based on legal knowledge graph and electronic equipment | |
| KR102119083B1 (en) | User review based rating re-calculation apparatus and method, storage media storing the same | |
| Chen et al. | The key successful factors of video and mobile game crowdfunding projects using a lexicon-based feature selection approach | |
| CN105931055A (en) | Service provider feature modeling method for crowdsourcing platform | |
| CN108595395B (en) | Nickname generation method, device and equipment | |
| Habbat et al. | LSTM-CNN deep learning model for French online product reviews classification | |
| CN112632275B (en) | Crowd clustering data processing method, device and equipment based on personal text information | |
| CN114239569A (en) | Analysis method and device for evaluation text and computer readable storage medium | |
| CN112115258B (en) | Credit evaluation method and device for user, server and storage medium | |
| CN114580427B (en) | Self-media user selection method and related equipment | |
| CN110851600A (en) | Text data processing method and device based on deep learning | |
| Sisodia et al. | Performance evaluation of learners for analyzing the hotel customer sentiments based on text reviews | |
| CN114896987B (en) | Fine-grained emotion analysis method and device based on semi-supervised pre-training model | |
| Hamza et al. | An Integrated Approach for Amazon Electronic Products Reviews by Using Sentiment Analysis | |
| Gadri et al. | An efficient system to predict customers’ satisfaction on touristic services using ML and DL approaches | |
| Jabeen | Decoding consumer sentiments: Advanced NLP techniques for analyzing smartphone reviews | |
| Zhang et al. | Learning sample representativeness for class-imbalanced multi-label classification |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |