CN114329184A - A method and system for expert recommendation in a question and answer website - Google Patents
A method and system for expert recommendation in a question and answer websiteDownload PDFInfo
- Publication number
- CN114329184A CN114329184ACN202111469719.6ACN202111469719ACN114329184ACN 114329184 ACN114329184 ACN 114329184ACN 202111469719 ACN202111469719 ACN 202111469719ACN 114329184 ACN114329184 ACN 114329184A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- answer
- question
- expert
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Translated fromChinese
Description
Translated fromChinese技术领域technical field
本发明涉及一种问答网站中专家推荐方法及系统，属于信息检索技术领域。The invention relates to an expert recommendation method and system in a question and answer website, belonging to the technical field of information retrieval.
背景技术Background technique
在知乎、Quora等问答网站中，找到与给定问题领域相关并愿意回答该问题的专家来提供高质量的回答不但能消除提问者的困惑，还能吸引更多用户参与讨论并加入感兴趣的社区，这使得网站的内容更加全面且提问者更容易得到满意的答案，对CQA(CommunityQuestion Answering)的发展非常重要。In Q&A sites such as Zhihu and Quora, finding experts who are relevant to a given problem domain and willing to answer the question to provide high-quality answers can not only eliminate the confusion of the questioner, but also attract more users to participate in the discussion and join interested The community, which makes the content of the website more comprehensive and the questioner easier to get satisfactory answers, is very important to the development of CQA (CommunityQuestion Answering).
现有技术中通常是通过对用户填写的认证内容的识别来区分该用户是不是专家，但实际应用中有很多用户不会填写认证内容，且用户填写的认证内容涉及的领域不全面，准确率低，导致上述领域专家发现的领域覆盖率低，且推荐的专家符合问题相应领域的准确度不高。In the prior art, whether the user is an expert is usually distinguished by identifying the authentication content filled in by the user, but in practical applications, many users do not fill in the authentication content, and the authentication content filled in by the user does not involve a comprehensive field, and the accuracy rate is high. Low, resulting in a low coverage rate of the fields discovered by experts in the above fields, and the accuracy of the recommended experts matching the corresponding fields of the problem is not high.
发明内容SUMMARY OF THE INVENTION
本发明的目的在于提供一种问答网站中专家推荐方法及系统，提高了专家推荐的准确度，降低了推荐计算的复杂度，提高了推荐的专家和指定问题的领域符合程度。The purpose of the present invention is to provide an expert recommendation method and system in a question and answer website, which improves the accuracy of expert recommendation, reduces the complexity of recommendation calculation, and improves the degree of conformity between the recommended expert and the specified problem.
为实现以上目的，本发明是采用下述技术方案实现的：To achieve the above object, the present invention adopts the following technical solutions to realize:
第一方面，本发明提供了一种问答网站中专家推荐方法，包括：In a first aspect, the present invention provides an expert recommendation method in a question and answer website, including:
获取全体用户的问答特征向量，将问答特征向量输入预建立的门控循环网络模型，得到时域语义特征向量；Obtain the question and answer feature vector of all users, input the question and answer feature vector into the pre-established gated recurrent network model, and obtain the time domain semantic feature vector;
根据用户的共同回答和关注关系构建共同回答-关注网络图；Build a common answer-follow network graph based on users’ common answers and attention relationships;
将时域语义特征向量输入预建立的图卷积神经网络中学习共同回答-关注网络图的拓扑结构，得到空域社交特征向量；Input the time-domain semantic feature vector into the pre-built graph convolutional neural network to learn the common answer-pay attention to the topology of the network graph, and obtain the spatial-domain social feature vector;
将空域社交特征向量输入全连接层进行分类，得到分类特征向量，分类特征向量中值为1的元素对应的用户为专家，得到专家列表；Input the airspace social feature vector into the fully connected layer for classification, and obtain the classification feature vector. The user corresponding to the element whose value is 1 in the classification feature vector is an expert, and the expert list is obtained;
获取专家的用户特征向量和指定问题特征向量，计算专家的用户特征向量和指定问题特征向量的余弦相似度，将余弦相似度大于预设阈值对应的专家作为推荐结果。Obtain the expert's user feature vector and the specified problem feature vector, calculate the cosine similarity between the expert's user feature vector and the specified problem feature vector, and use the expert whose cosine similarity is greater than the preset threshold as the recommendation result.
结合第一方面，进一步的，所述全体用户的问答特征向量通过以下方式得到：In combination with the first aspect, further, the question-and-answer feature vectors of all users are obtained in the following manner:
其中
结合第一方面，进一步的，所述共同回答-关注网络图通过以下方法构建：In combination with the first aspect, further, the common answer-attention network graph is constructed by the following methods:
在问答网站中任意两个用户共同回答了同一个问题或某个用户关注了另一用户，则这两个用户之间就有一条边相连，据此构建图G＝<V,E>，图G＝<V,E>为共同回答-关注网络图；其中，V为用户集，E为边集。In the question-and-answer website, if any two users answer the same question together or a user follows another user, there is an edge between the two users. According to this, a graph G=<V, E> is constructed, and the graph G=<V, E> is the common answer-attention network graph; where V is the user set and E is the edge set.
结合第一方面，进一步的，将问答特征向量输入预建立的门控循环网络模型，得到时域语义特征向量：Combined with the first aspect, further, the question and answer feature vector is input into the pre-established gated recurrent network model to obtain the time-domain semantic feature vector:
ht＝ut*ht-1+(1-ut)*ctht =ut *ht-1 +(1-ut )*ct
其中，ht-1为t-1时刻门控循环网络模型的输出，ht为t时刻门控循环网络模型的输出，门控循环网络模型最终的输出为时域语义特征向量，σ表示第一非线性激活函数，Wu表示更新门权重矩阵，
结合第一方面，进一步的，将时域语义特征向量输入预建立的图卷积神经网络中学习共同回答-关注网络图的拓扑结构，得到空域社交特征向量：Combined with the first aspect, further, the time-domain semantic feature vector is input into the pre-established graph convolutional neural network to learn the common answer-pay attention to the topology of the network graph, and the spatial-domain social feature vector is obtained:
H0＝LH0 =L
其中，L为时域语义特征向量，H0代表图卷积神经网络的输入，Hl+1代表第l层图卷积神经网络的输出，σ表示第一非线性激活函数，
结合第一方面，进一步的，将空域社交特征向量输入全连接层进行分类，得到分类特征向量：Combined with the first aspect, further, the spatial social feature vector is input into the fully connected layer for classification, and the classification feature vector is obtained:
YU＝softmax(δ(W2Z+b2))YU =softmax(δ(W2 Z+b2 ))
其中，δ表示第二非线性激活函数，W1代表第一层全连接层的可训练权重矩阵，XU是用户的特征向量，H是空域社交特征向量，b1是第一层全连接层的偏置向量，Z是第一层全连接层的输出，W2代表第二层全连接层的可训练权重矩阵，b2是第二层全连接层的偏置向量，YU是分类特征向量。where δ represents the second nonlinear activation function, W1 represents the trainable weight matrix of the first fully connected layer, XU is the feature vector of the user, H is the spatial social feature vector, and b1 is the first fully connected layer , Z is the output of the first fully connected layer, W2 represents the trainable weight matrix of thesecond fully connected layer, b2 is the bias vector of thesecond fully connected layer, YU is the classification feature vector.
结合第一方面，进一步的，还包括将满足预设要求的用户标记为专家的步骤：Combined with the first aspect, further, it further includes the step of marking users who meet the preset requirements as experts:
若用户某一周的点赞数大于预设点赞数阈值，则该用户在该周被标记为专家，若用户在某一周未回答问题，则沿用其上周的点赞数。If the number of likes of a user in a certain week is greater than the preset threshold of likes, the user will be marked as an expert in that week, and if the user does not answer a question in a certain week, the likes of the previous week will be used.
第二方面，本发明还提供了一种问答网站中专家推荐系统，包括：In the second aspect, the present invention also provides an expert recommendation system in a question and answer website, including:
时域特征提取模块：用于获取全体用户的问答特征向量，将问答特征向量输入预建立的门控循环网络模型，得到时域语义特征向量；Time domain feature extraction module: used to obtain the question and answer feature vector of all users, input the question and answer feature vector into the pre-established gated recurrent network model, and obtain the time domain semantic feature vector;
网络图构建模块：用于根据用户的共同回答和关注关系构建共同回答-关注网络图；Network graph building module: used to construct a common answer-attention network graph based on users’ common answers and attention relationships;
空域特征提取模块：用于将时域语义特征向量输入预建立的图卷积神经网络中学习共同回答-关注网络图的拓扑结构，得到空域社交特征向量；Spatial domain feature extraction module: used to input the time domain semantic feature vector into the pre-established graph convolutional neural network to learn the common answer-pay attention to the topology of the network graph, and obtain the spatial domain social feature vector;
专家提取模块：用于将空域社交特征向量输入全连接层进行分类，得到分类特征向量，分类特征向量中值为1的元素对应的用户为专家，得到专家列表；Expert extraction module: It is used to input the spatial social feature vector into the fully connected layer for classification, and obtain the classification feature vector. The user corresponding to the element whose value is 1 in the classification feature vector is an expert, and the expert list is obtained;
推荐模块：用于获取专家的用户特征向量和指定问题特征向量，计算专家的用户特征向量和指定问题特征向量的余弦相似度，将余弦相似度大于预设阈值对应的专家作为推荐结果。Recommendation module: used to obtain the expert's user feature vector and the specified problem feature vector, calculate the cosine similarity between the expert's user feature vector and the specified problem feature vector, and use the expert whose cosine similarity is greater than the preset threshold as the recommendation result.
与现有技术相比，本发明所达到的有益效果是：Compared with the prior art, the beneficial effects achieved by the present invention are:
本发明提供的一种问答网站中专家推荐方法及系统，将问答特征向量输入预建立的门控循环网络模型，使用门控循环网络模型学习随时间动态变化的问答特征向量以捕获时域语义特征向量，考虑到了每位用户的知识储备和专业技能会随时间动态增长的实际状况，本发明结合了此实际情况，提高了专家推荐的准确度；根据用户的共同回答和关注关系构建共同回答-关注网络图，将专家发现问题转变为图节点分类问题，降低了推荐计算的复杂度；通过计算专家的用户特征向量和指定问题特征向量的余弦相似度，将余弦相似度大于预设阈值对应的专家作为推荐结果，提高了推荐的专家和指定问题的领域符合程度。The present invention provides a method and system for expert recommendation in a question-and-answer website. The question-and-answer feature vector is input into a pre-established gated cyclic network model, and the gated cyclic network model is used to learn the question-and-answer feature vector that changes dynamically with time to capture time-domain semantic features. vector, taking into account the actual situation that each user's knowledge reserve and professional skills will dynamically increase over time, the present invention combines this actual situation to improve the accuracy of expert recommendation; build a common answer according to the user's common answer and concern relationship- Pay attention to the network graph, transform the problem of expert discovery into the problem of graph node classification, and reduce the complexity of recommendation calculation; by calculating the cosine similarity between the expert's user feature vector and the specified problem feature vector, the cosine similarity greater than the preset threshold corresponds to Experts are recommended as a result, which improves the degree of domain conformity between the recommended experts and the specified problem.
附图说明Description of drawings
图1是本发明实施例提供的一种问答网站中专家推荐方法的流程图；Fig. 1 is a flowchart of an expert recommendation method in a question-and-answer website provided by an embodiment of the present invention;
图2是本发明实施例提供的共同回答-关注网络图；FIG. 2 is a common answer-attention network diagram provided by an embodiment of the present invention;
图3是本发明实施例提供的GRU-GCN模型的示意图。FIG. 3 is a schematic diagram of a GRU-GCN model provided by an embodiment of the present invention.
具体实施方式Detailed ways
下面结合附图对本发明作进一步描述，以下实施例仅用于更加清楚地说明本发明的技术方案，而不能以此来限制本发明的保护范围。The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and cannot be used to limit the protection scope of the present invention.
实施例1Example 1
如图1所示，本发明实施例提供了一种问答网站中专家推荐方法，包括：As shown in FIG. 1 , an embodiment of the present invention provides a method for recommending experts in a question-and-answer website, including:
获取全体用户的问答特征向量，将问答特征向量输入预建立的门控循环网络模型，得到时域语义特征向量。The question and answer feature vectors of all users are obtained, and the question and answer feature vectors are input into the pre-established gated recurrent network model to obtain the time-domain semantic feature vectors.
从问答网站获取脱敏后的用户信息及用户问答信息并保存到服务器，对数据进行预处理，即对问答文本做数据清洗，清除无关字符。Obtain the desensitized user information and user question and answer information from the question and answer website and save it to the server, and preprocess the data, that is, perform data cleaning on the question and answer text to remove irrelevant characters.
在专家发现的过程中，需要将用户的问答文本经Bert转换为问答特征向量，而问答文本中会包含部分代码片段、公式和一些无用的符号，这将成为问答特征向量的噪声，所以需要从问答文本中提取出真正有用的信息，使用正则表达式对问答文本进行过滤。In the process of expert discovery, the user's question and answer text needs to be converted into question and answer feature vector by Bert, and the question and answer text will contain some code fragments, formulas and some useless symbols, which will become the noise of the question and answer feature vector, so it needs to be extracted from the question and answer text. The really useful information is extracted from the question and answer text, and the question and answer text is filtered using regular expressions.
构造问答特征向量，对于身为回答者的用户，记录其每周提供的问答文本，如果其在某一周未回答，则使用其上一周的问答文本，将用户的问答文本经Bert处理，如图3所示，全体用户的问答特征向量可以表示为：Construct the question and answer feature vector. For the user who is the answerer, record the question and answer text provided every week. If he does not answer in a certain week, the question and answer text of the previous week is used, and the user's question and answer text is processed by Bert, as shown in the figure. 3, the question-answer feature vector of all users can be expressed as:
其中
还包括标记专家的步骤，通过用户每周的点赞数来标记专家，若用户某一周的点赞数大于预设点赞数阈值，则该用户在该周被标记为专家，若用户在某一周未回答问题，则沿用其上周的点赞数。It also includes the step of marking experts. The experts are marked by the user's weekly likes. If the user's likes in a certain week is greater than the preset likes threshold, the user is marked as an expert in that week. If the question is not answered for a week, the number of likes from the previous week will be used.
将问答特征向量放入预建立的门控循环网络模型(GRU)学习随时间动态变化的问答特征向量以捕获时域语义特征向量：The question and answer feature vectors are put into a pre-built gated recurrent network model (GRU) to learn the question and answer feature vectors that change dynamically over time to capture the temporal semantic feature vectors:
ht＝ut*ht-1+(1-ut)*ctht =ut *ht-1 +(1-ut )*ct
其中，ht-1为t-1时刻门控循环网络模型的输出，ht为t时刻门控循环网络模型的输出，门控循环网络模型最终的输出为时域语义特征向量，σ表示第一非线性激活函数，Wu表示更新门权重矩阵，
根据用户的共同回答和关注关系构建共同回答-关注网络图。Construct a co-answer-attention network graph based on users' co-answer and attention relationships.
如图2所示，在问答网站中，任意两个用户共同回答了同一个问题或某个用户关注了另一用户，则这两个用户之间就有一条边相连，据此构建图G＝<V,E>，图G＝<V,E>为共同回答-关注网络图；其中，G是一个无向图，V为用户集，E为边集。As shown in Figure 2, in the question-and-answer website, if any two users answer the same question together or a user follows another user, there is an edge between the two users, and a graph G = <V,E>, the graph G=<V,E> is the common answer-attention network graph; where G is an undirected graph, V is the user set, and E is the edge set.
将时域语义特征向量输入预建立的图卷积神经网络中学习共同回答-关注网络图的拓扑结构，得到空域社交特征向量。The temporal semantic feature vector is input into a pre-built graph convolutional neural network to learn the common answer-attention network graph topology, and the spatial social feature vector is obtained.
如图3所示，将GRU的输出作为初始向量放入预建立的图卷积神经网络(GCN)中，将图G＝<V,E>的邻接矩阵也一起放入GCN中从网络中提取信息，通过堆叠GCN层，最后一层GCN的输出H为空域社交特征向量，H不仅包含节点自己的特征，还包含自己邻居的特征。As shown in Figure 3, the output of the GRU is put into the pre-established graph convolutional neural network (GCN) as the initial vector, and the adjacency matrix of the graph G=<V, E> is also put into the GCN and extracted from the network. Information, by stacking GCN layers, the output H of the last layer of GCN is the spatial social feature vector, H contains not only the features of the node itself, but also the features of its own neighbors.
H0＝LH0 =L
其中，L为时域语义特征向量，H0代表图卷积神经网络的输入，Hl+1代表第l层图卷积神经网络的输出，σ表示第一非线性激活函数，
将空域社交特征向量输入全连接层进行分类，得到分类特征向量，分类特征向量中值为1的元素对应的用户为专家，得到专家列表。The spatial social feature vector is input into the fully connected layer for classification, and the classification feature vector is obtained. The user corresponding to the element whose value is 1 in the classification feature vector is an expert, and an expert list is obtained.
YU＝softmax(δ(W2Z+b2))YU =softmax(δ(W2 Z+b2 ))
其中，δ表示第二非线性激活函数，W1代表第一层全连接层的可训练权重矩阵，XU是用户的特征向量，图3中XU1至XU6表示第1至第6个用户的特征向量，H是空域社交特征向量，b1是第一层全连接层的偏置向量，Z是第一层全连接层的输出，W2代表第二层全连接层的可训练权重矩阵，b2是第二层全连接层的偏置向量，YU是分类特征向量。Among them, δ represents the second nonlinear activation function, W1 represents the trainable weight matrix of the first fully connected layer, XU is the feature vector of the user, and XU1 to XU6 in Figure 3 represent the first to sixth users , H is the spatial social feature vector, b1 is the bias vector of thefirst fully connected layer, Z is the output of the first fully connected layer, W2 represents the trainable weight matrix of thesecond fully connected layer , b2 is the bias vector of the second fully connected layer, and YU is the categorical feature vector.
XU是用户的特征向量，对应着用户的用户资料，用户资料包括五种属性，分别是粉丝数、用户简介、是否是机构用户、是否是广告用户、是否是匿名用户，每种属性对应一个一维向量，这样用户资料对应的用户的特征向量长度为5，即XU∈Rn×5，其中n为用户总数。XU is the feature vector of the user, corresponding to the user's user profile. The user profile includes five attributes, namely the number of fans, user profile, whether it is an institutional user, whether it is an advertising user, and whether it is an anonymous user. Each attribute corresponds to one A one-dimensional vector, so that the length of the feature vector of the user corresponding to the user profile is 5, that is, XU ∈ Rn×5 , where n is the total number of users.
YU中第i位元素为1时，说明第i个用户为专家，从而得到专家列表。When the i-th element in YU is 1, it means that the i-th user is an expert, thus obtaining an expert list.
获取专家的用户特征向量和指定问题特征向量，计算专家的用户特征向量和指定问题特征向量的余弦相似度，将余弦相似度大于预设阈值对应的专家作为推荐结果。Obtain the expert's user feature vector and the specified problem feature vector, calculate the cosine similarity between the expert's user feature vector and the specified problem feature vector, and use the expert whose cosine similarity is greater than the preset threshold as the recommendation result.
当有问题需要推荐专家时，将指定问题的文本经Bert处理得到指定问题特征，获取专家的用户特征向量XU和指定问题特征向量，计算专家的用户特征向量和指定问题特征向量的余弦相似度，将余弦相似度大于预设阈值对应的专家作为推荐结果，也可以将余弦相似度大于预设阈值的前n名对应的专家作为推荐结果。When there is a problem to recommend an expert, the text of the specified problem is processed by Bert to obtain the specified problem feature, the user feature vector XU of the expert and the specified problem feature vector are obtained, and the cosine similarity between the expert user feature vector and the specified problem feature vector is calculated. , the experts whose cosine similarity is greater than the preset threshold are used as the recommendation result, and the top n experts whose cosine similarity is greater than the preset threshold may also be used as the recommendation result.
实施例2Example 2
本发明实施例提供了一种问答网站中专家推荐系统，包括：The embodiment of the present invention provides an expert recommendation system in a question and answer website, including:
时域特征提取模块：用于获取全体用户的问答特征向量，将问答特征向量输入预建立的门控循环网络模型，得到时域语义特征向量；Time domain feature extraction module: used to obtain the question and answer feature vector of all users, input the question and answer feature vector into the pre-established gated recurrent network model, and obtain the time domain semantic feature vector;
网络图构建模块：用于根据用户的共同回答和关注关系构建共同回答-关注网络图；Network graph building module: used to construct a common answer-attention network graph based on users’ common answers and attention relationships;
空域特征提取模块：用于将时域语义特征向量输入预建立的图卷积神经网络中学习共同回答-关注网络图的拓扑结构，得到空域社交特征向量；Spatial domain feature extraction module: used to input the time domain semantic feature vector into the pre-established graph convolutional neural network to learn the common answer-pay attention to the topology of the network graph, and obtain the spatial domain social feature vector;
专家提取模块：用于将空域社交特征向量输入全连接层进行分类，得到分类特征向量，分类特征向量中值为1的元素对应的用户为专家，得到专家列表；Expert extraction module: It is used to input the spatial social feature vector into the fully connected layer for classification, and obtain the classification feature vector. The user corresponding to the element whose value is 1 in the classification feature vector is an expert, and the expert list is obtained;
推荐模块：用于获取专家的用户特征向量和指定问题特征向量，计算专家的用户特征向量和指定问题特征向量的余弦相似度，将余弦相似度大于预设阈值对应的专家作为推荐结果。Recommendation module: used to obtain the expert's user feature vector and the specified problem feature vector, calculate the cosine similarity between the expert's user feature vector and the specified problem feature vector, and use the expert whose cosine similarity is greater than the preset threshold as the recommendation result.
实施例3Example 3
本发明实施例提供了一种问答网站中专家推荐系统，包括：The embodiment of the present invention provides an expert recommendation system in a question and answer website, including:
数据抽取模块：用于利用爬虫从社区问答网站不断爬取所需信息，比如所有用户的问答文本，并抽取相关数据及对数据进行处理。Data extraction module: It is used to use crawlers to continuously crawl the required information from the community question and answer website, such as the question and answer texts of all users, and extract relevant data and process the data.
存储模块：用于对爬取的信息进行有效存储，本实施方案采用sqlite进行数据存储，所有对数据库操作都通过调用接口实现。Storage module: used to effectively store the crawled information. In this embodiment, sqlite is used for data storage, and all operations on the database are implemented through the calling interface.
算法模块：用于使用实施例1所述的一种问答网站中专家推荐方法，发现专家并将专家的用户特征向量和指定问题特征向量的余弦相似度大于预设阈值对应的专家作为推荐结果。Algorithm module: used to use the method for recommending experts in a question-and-answer website described in
服务模块：用于将该系统所实现的功能以Web服务的方式提供给用户。Service module: used to provide the functions implemented by the system to users in the form of Web services.
本领域内的技术人员应明白，本申请的实施例可提供为方法、系统、或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。As will be appreciated by those skilled in the art, the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
本申请是参照根据本申请实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It will be understood that each process and/or block in the flowchart illustrations and/or block diagrams, and combinations of processes and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device produce Means for implementing the functions specified in a flow or flow of a flowchart and/or a block or blocks of a block diagram.
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions The apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。These computer program instructions can also be loaded on a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that The instructions provide steps for implementing the functions specified in the flow or blocks of the flowcharts and/or the block or blocks of the block diagrams.
以上所述仅是本发明的优选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本发明技术原理的前提下，还可以做出若干改进和变形，这些改进和变形也应视为本发明的保护范围。The above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the technical principle of the present invention, several improvements and modifications can also be made. These improvements and modifications It should also be regarded as the protection scope of the present invention.
Claims (8)
Translated fromChinese
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111469719.6ACN114329184A (en) | 2021-12-03 | 2021-12-03 | A method and system for expert recommendation in a question and answer website |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111469719.6ACN114329184A (en) | 2021-12-03 | 2021-12-03 | A method and system for expert recommendation in a question and answer website |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114329184Atrue CN114329184A (en) | 2022-04-12 |
Family
ID=81048280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111469719.6APendingCN114329184A (en) | 2021-12-03 | 2021-12-03 | A method and system for expert recommendation in a question and answer website |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114329184A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114896384A (en)* | 2022-06-14 | 2022-08-12 | 青岛科技大学 | Question answering expert cross-domain recommendation method and system for question answering community |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108021616A (en)* | 2017-11-06 | 2018-05-11 | 大连理工大学 | A community question answering expert recommendation method based on recurrent neural network |
| CN110321421A (en)* | 2019-07-04 | 2019-10-11 | 南京邮电大学 | Expert recommendation method and computer storage medium for website Knowledge Community system |
| US20200167428A1 (en)* | 2018-11-26 | 2020-05-28 | International Business Machines Corporation | Utilizing external knowledge and memory networks in a question-answering system |
| CN113256438A (en)* | 2021-05-27 | 2021-08-13 | 山东省计算中心（国家超级计算济南中心） | Role identification method and system for network user |
- 2021
- 2021-12-03CNCN202111469719.6Apatent/CN114329184A/enactivePending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108021616A (en)* | 2017-11-06 | 2018-05-11 | 大连理工大学 | A community question answering expert recommendation method based on recurrent neural network |
| US20200167428A1 (en)* | 2018-11-26 | 2020-05-28 | International Business Machines Corporation | Utilizing external knowledge and memory networks in a question-answering system |
| CN110321421A (en)* | 2019-07-04 | 2019-10-11 | 南京邮电大学 | Expert recommendation method and computer storage medium for website Knowledge Community system |
| CN113256438A (en)* | 2021-05-27 | 2021-08-13 | 山东省计算中心（国家超级计算济南中心） | Role identification method and system for network user |
Non-Patent Citations (2)
| Title |
|---|
| JINFENG XIAO 等: "Open-Domain Question Answering with Pre-Constructed Question Spaces", PROCEEDINGS OF NAACL-HLT 2021, 11 June 2021 (2021-06-11), pages 1 - 7* |
| 刘臣 等: "基于图卷积神经网络的在线社区行为预测", 计算机技术与发展, vol. 31, no. 4, 30 April 2021 (2021-04-30), pages 1 - 7* |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114896384A (en)* | 2022-06-14 | 2022-08-12 | 青岛科技大学 | Question answering expert cross-domain recommendation method and system for question answering community |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111931062B (en) | Training method and related device of information recommendation model | |
| CN112307351B (en) | User behavior model training, recommendation methods, devices and equipment | |
| Riederer et al. | Linking users across domains with location data: Theory and validation | |
| US8590049B2 (en) | Method and apparatus for providing anonymization of data | |
| US20110275047A1 (en) | Seeking Answers to Questions | |
| CN109062914B (en) | User recommendation method and device, storage medium and server | |
| CN110321421B (en) | Expert recommendation method for website knowledge community system and computer storage medium | |
| CN113158077B (en) | A method for recommending academic resources based on user portraits | |
| CN109903127A (en) | Group recommendation method and device, storage medium and server | |
| JP2019113943A (en) | Information providing apparatus, information providing method, and program | |
| WO2021155691A1 (en) | User portrait generating method and apparatus, storage medium, and device | |
| CN115408621B (en) | Point-of-interest recommendation method considering linear and nonlinear interaction of auxiliary information features | |
| CN114116995A (en) | Conversational recommendation method, system and medium based on augmented graph neural network | |
| Alabduljabbar et al. | A dynamic selection approach for quality control mechanisms in crowdsourcing | |
| CN107103499A (en) | Method is recommended in a kind of cross-site cold start-up based on deep learning | |
| CN113422761A (en) | Malicious social user detection method based on counterstudy | |
| WO2023155306A1 (en) | Data recommendation method and apparatus based on graph neural network and electronic device | |
| CN118013022A (en) | A social media data aggregation analysis system and method | |
| CN114329184A (en) | A method and system for expert recommendation in a question and answer website | |
| US20180121824A1 (en) | Artificial Intelligence for Decision Making Based on Machine Learning of Human Decision Making Process | |
| Liu et al. | Identifying experts in community question answering website based on graph convolutional neural network | |
| CN111353001A (en) | Method and apparatus for classifying users | |
| Cuomo et al. | A biologically inspired model for describing the user behaviors in a Cultural Heritage environment. | |
| CN105373591A (en) | Construction method of user interest model based on temporal-spatial behavior pattern in mobile social network | |
| Singh et al. | Patterns affecting structural properties of social networking site'Twitter' |
Legal Events
| 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 |