技术领域technical field
本发明涉及一种专家推荐方法及计算机存储介质，特别是涉及一种用于网站知识社区系统的专家推荐方法及计算机存储介质。The invention relates to an expert recommendation method and a computer storage medium, in particular to an expert recommendation method and a computer storage medium for a website knowledge community system.
背景技术Background technique
知识问答社区是一种新的Web应用，用户可以通过在社区中以提问和回答问题的方式进行知识交流。在知识问答社区诞生之前，人们主要使用搜索引擎主动的从网上获取信息。搜索引擎的核心在于关键字匹配，但是由于关键字属于短文本，基本上无法对其进行语义分析，从而导致大量的搜索结果与用户的意图存在偏差。并且搜索引擎返回的结果太多，用户很难快速准确的找到自己所需信息。而在知识问答社区中，用户通过以自然语言的形式提出具体详细的问题并且可以从其他用户那里获取答案来满足自己的需求。但是，随着社区中用户数量指数级的增长，新问题的潜在回答者往往需要花费大量的时间和精力来寻找自己感兴趣并且有能力回答的问题，社区中新问题的提问者需要花费几个小时甚至几天的时间来等待问题被回答。目前，大部分的知识问答社区都存在着一个普遍的现象：有很多问题长期无人回复并且这些问题的数量与日俱增。Knowledge Q&A community is a new web application, users can exchange knowledge by asking and answering questions in the community. Before the birth of the knowledge question and answer community, people mainly used search engines to actively obtain information from the Internet. The core of search engines is keyword matching, but because keywords belong to short texts, it is basically impossible to perform semantic analysis on them, resulting in a large number of search results that deviate from users' intentions. And search engines return too many results, making it difficult for users to quickly and accurately find the information they need. In the knowledge question answering community, users meet their own needs by asking specific and detailed questions in natural language and getting answers from other users. However, with the exponential growth of the number of users in the community, potential answerers of new questions often need to spend a lot of time and energy to find questions that they are interested in and capable of answering. Hours or even days waiting for questions to be answered. At present, there is a common phenomenon in most of the knowledge Q&A communities: there are many questions that go unanswered for a long time and the number of these questions is increasing day by day.
运用面向网站知识社区系统的专家推荐方法是解决上述问题的有效途径，目前的方法主要分为三种：第一种是根据社区中用户之间的交互关系构建用户问答关系有向图，然后使用基于图的排序算法实现专家推荐。第二种是利用社区中丰富的文本特征数据，利用主题模型及其优化模型对文本进行语义分析。第三种是将基于图的排序算法和基于文本的语义分析技术联合使用。这些方法虽然取得了一定的效果，但是普遍存在以下几个问题：Using the expert recommendation method for the website knowledge community system is an effective way to solve the above problems. The current methods are mainly divided into three types. Graph-based sorting algorithm for expert recommendation. The second is to use the rich text feature data in the community to perform semantic analysis on the text using the topic model and its optimization model. The third is to combine graph-based sorting algorithms and text-based semantic analysis techniques. Although these methods have achieved certain results, there are generally the following problems:
1、对传统的基于图的排序算法进行改进时忽略了用户回答问题的平均响应时间，从而导致了推荐的专家用户提供回复不够及时；1. The improvement of the traditional graph-based sorting algorithm ignores the average response time of users answering questions, which leads to the lack of timely responses provided by the recommended expert users;
2、在对社区中文本进行语义分析大多采用隐含狄利克雷分布(Latent DirichletAllocation，LDA)模型及其优化模型，但这些模型的泛化能力比较弱，严重影响了推荐结果的准确性；2. The latent Dirichlet Allocation (LDA) model and its optimization model are mostly used in the semantic analysis of the text in the community, but the generalization ability of these models is relatively weak, which seriously affects the accuracy of the recommendation results;
3、在构建随机概率转移矩阵的权重时，考虑的因素过于单一，导致推荐的专家权威性有限。3. When constructing the weight of the random probability transition matrix, the factors considered are too single, resulting in limited authority of the recommended experts.
发明内容SUMMARY OF THE INVENTION
发明目的：本发明要解决的技术问题是提供一种用于网站知识社区系统的专家推荐方法及计算机存储介质，解决了目前的推荐方法准确性不高、推荐的专家权威性有限和专家提供回复不及时等不足，该方法推荐的专家不仅具有较高的兴趣度和专长度，而且还有一定的权威性。Purpose of the invention: The technical problem to be solved by the present invention is to provide an expert recommendation method and a computer storage medium for a website knowledge community system, which solves the problem that the current recommendation method is not accurate, the authority of the recommended experts is limited, and the experts provide replies. If it is not timely enough, the experts recommended by this method not only have a high degree of interest and expertise, but also have a certain degree of authority.
技术方案：本发明所述的用于网站知识社区系统的专家推荐方法，包括以下步骤：Technical solution: the expert recommendation method for the website knowledge community system according to the present invention comprises the following steps:
(1)获取知识问答社区的数据集并进行预处理，；(1) Obtain the data set of the knowledge question answering community and preprocess it;
(2)将数据集中各用户提问的问题和回答的答案合并为用户文本；(2) Combine the questions asked and answered by each user in the dataset into user texts;
(3)建立深度结构化语义模型并进行训练，使用训练好的深度结构化语义模型计算问题文本和用户文本之间的相关度，根据相关度的大小筛选出候选专家群；(3) Establish a deep structured semantic model and train it, use the trained deep structured semantic model to calculate the correlation between the question text and the user text, and screen out candidate expert groups according to the correlation;
(4)针对步骤(3)中的候选专家群，构建用户问答关系有向图；(4) For the candidate expert group in step (3), construct a directed graph of the user question and answer relationship;
(5)针对步骤(4)中的用户问答关系有向图运用排序算法进行链接分析，计算各用户的权威度值，选出权威度值最高的TOP-N个结果。(5) Use a sorting algorithm to perform link analysis on the directed graph of the user question and answer relationship in step (4), calculate the authority value of each user, and select the TOP-N results with the highest authority value.
进一步的，步骤(5)中的计算各用户的权威度值具体为：Further, calculating the authority value of each user in step (5) is specifically:
其中，A(uj)表示用户j的权威度值，A(ui)表示用户ui的权威度值，β是阻尼因子，N′表示构造的有向图中节点的个数，e表示从用户ui出发到用户uj构成的有向边，表示用户ui到用户uj的权重，∑ui表示用户ui到所有用户的权重和。Among them, A(uj ) represents the authority value of user j, A(ui ) represents the authority value of userui , β is the damping factor, N′ represents the number of nodes in the constructed directed graph, and e represents From user ui to the directed edge formed by user uj , represents the weight from user ui to user uj , and ∑ui represents the sum of the weights from userui to all users.
进一步的，所述权重的公式为：Further, the weight The formula is:
其中，Nij表示用户uj累计回答用户ui问题的个数，表示用户uj回答用户ui提出的问题的难度值，Tavg表示用户uj对用户ui提出的问题的平均响应时间。Among them, Nij represents the cumulative number of user uj answers to userui ’s questions, Represents the difficulty value of user uj answering the question posed by userui , and Tavg represents the average response time of user uj to the question posed by userui .
进一步的，所述难度值的公式为：Further, the difficulty value The formula is:
其中，表示用户uk回答用户ui问题的时刻，表示用户ui提出问题的时刻，N表示用户ui提出的问题获得的答案总数。in, represents the moment when useruk answers the question of userui , Represents the moment when userui asks the question, and N represents the total number of answers to the question raised by userui .
进一步的，所述平均响应时间Tavg的公式为：Further, the formula of the average response time Tavg is:
其中，Tjk表示用户uj回答用户ui提出的第k个问题的时刻，Tik表示用户ui提出第k个问题的时刻，Nij表示用户uj回答用户ui提出的问题总数。Among them, Tjk represents the moment when user uj answers the kth question raised by userui , Tik represents the moment when userui asks the kth question, and Nij represents the total number of questions that user uj answers to userui .
进一步的，构建用户问答关系有向图的方法为：从所有提出问题的用户出发建立链接，链接方向指向为该问题提供答案的用户。Further, the method for constructing a directed graph of user question-answer relationships is to establish links from all users who ask questions, and the link direction points to users who provide answers to the question.
本发明所述的计算机存储介质，其上存储有计算机程序，所述计算机程序在被计算机处理器执行时实现上述用于网站知识社区系统的专家推荐方法。The computer storage medium of the present invention stores a computer program thereon, and when the computer program is executed by a computer processor, implements the above-mentioned expert recommendation method for a website knowledge community system.
有益效果：本发明具有以下技术效果：Beneficial effects: the present invention has the following technical effects:
1、本发明在对问题文本和用户文本进行语义分析时，运用了深度结构化语义分析模型。在使用深度结构化语义分析模型后，运用主题敏感性回答者排序算法对由候选专家构建的用户问答关系有向图进行链接分析。本专家推荐方法具有很好的普适性，在不同的知识问答社区均可实现精准的专家推荐。1. The present invention uses a deep structured semantic analysis model when performing semantic analysis on question text and user text. After using a deep structured semantic analysis model, a topic-sensitive answerer ranking algorithm is used to perform link analysis on the directed graph of user question-answer relationships constructed by candidate experts. This expert recommendation method has good universality, and can achieve accurate expert recommendation in different knowledge question and answer communities.
2、本发明专家推荐方法可以统一部署在服务器端，根据用户的需求在不同的客户端进行专家推荐结果展示。2. The expert recommendation method of the present invention can be uniformly deployed on the server side, and the expert recommendation results can be displayed on different clients according to the needs of the user.
3、本发明专家推荐方法推荐的候选专家不仅具有一定的权威性，而且对具体的问题也具备较高的专业性，有效的改善了知识问答社区的用户体验，提高了用户满意度。3. The candidate experts recommended by the expert recommendation method of the present invention not only have certain authority, but also have high professionalism for specific problems, which effectively improves the user experience of the knowledge question and answer community and improves user satisfaction.
附图说明Description of drawings
图1是本发明实施方式方法的流程图；Fig. 1 is the flow chart of the embodiment method of the present invention;
图2是本发明实施方式中深度结构化语义模型训练流程图；Fig. 2 is the deep structured semantic model training flow chart in the embodiment of the present invention;
图3是本发明实施方式中用户问答关系有向图。FIG. 3 is a directed graph of a question-answer relationship between users in an embodiment of the present invention.
具体实施方式Detailed ways
本发明提出的用于网站知识社区系统的专家推荐方法，首先使用深度结构化语义模型进行文本分析，然后使用主题敏感性专家排序算法对用户问答关系有向图进行深入的链接分析，最后对链接分析的结果进行降序排序并生成推荐专家列表。具体的实施方式如图1所示，包括以下步骤：The expert recommendation method for the website knowledge community system proposed by the present invention firstly uses a deep structured semantic model to perform text analysis, then uses a topic-sensitive expert sorting algorithm to perform in-depth link analysis on the directed graph of user question-answer relationships, and finally performs an in-depth link analysis on the links. The results of the analysis are sorted in descending order and a list of recommended experts is generated. The specific embodiment, as shown in Figure 1, includes the following steps:
步骤1，首先整体获取知识问答社区(如：Yahoo！Answers,Stack Overflow)的数据集并对其进行预处理。预处理主要包括以下几个子步骤：1.1处理原始数据集中冗余的信息，去除文本中的停用词(Stop Words)以及标点符号；1.2进一步对文本进行分割，将其分割为词或字母并转换成向量，抽取词或字母的元组，再将每个元组转换成向量。Step 1, first obtain the dataset of knowledge question answering community (eg: Yahoo! Answers, Stack Overflow) as a whole and preprocess it. Preprocessing mainly includes the following sub-steps: 1.1 Process the redundant information in the original data set, remove stop words (Stop Words) and punctuation marks in the text; 1.2 Further segment the text into words or letters and convert them into a vector, extract tuples of words or letters, and convert each tuple into a vector.
然后将分散的用户交互历史行为数据以问题-答案为中心进行重新整理，将用户在社区中新提出的问题标记为问题文本，将用户曾经提问的问题和回答的答案合并成一个文本并标记为用户文本。Then the scattered user interaction historical behavior data is reorganized around the question-answer center, the new questions raised by users in the community are marked as question text, the questions that users have asked and the answers answered are combined into one text and marked as User text.
步骤2，对深度结构化语义模型(DSSM)进行训练，如图2所示，按照如下步骤2.1至2.4，然后利用训练好的模型计算问题文本和用户文本之间的相关性，根据相关性值的大小生成候选专家群。Step 2, train the Deep Structured Semantic Model (DSSM), as shown in Figure 2, follow steps 2.1 to 2.4 below, and then use the trained model to calculate the correlation between the question text and the user text, according to the correlation value to generate candidate expert groups.
步骤2.1定义损失函数。通过极大似然估计，最小化损失函数，公式如下：Step 2.1 Define the loss function. Through maximum likelihood estimation, the loss function is minimized, and the formula is as follows:
其中，Uk是提供答案的专家集合，Q表示社区中用户提出的问题，Pr表示条件概率，Wi表示第i层的权值矩阵，bi表示第i层的偏差。Among them, Uk is the set of experts who provide answers, Q is the question raised by users in the community, Pr is the conditional probability, Wi is the weight matrix of thei -th layer, and bi is the bias of the i-th layer.
步骤2.2反向传播，更新权值参数。在表示层的深度神经网络中，残差进行反向传播，运用随机梯度下降算法使模型收敛，得到各网络层的参数{wi,bi}，获取文本语义特征的公式为：Step 2.2 Backpropagation, update the weight parameters. In the deep neural network of the presentation layer, the residuals are back-propagated, and the stochastic gradient descent algorithm is used to make the model converge to obtain the parameters {wi ,bi } of each network layer. The formula for obtaining the semantic features of the text is:
其中，xhi(k)表示在时刻K哈希后的输入值，i从0到m，Wi(k)表示在时刻k的权重值，b是偏差值，F是激活函数，y(k)表示在时刻Wi(k)的输出值。隐藏层和输出层之间的激活函数为：Among them, xhi (k) represents the input value after hashing at time K,i ranges from 0 to m, Wi (k) represents the weight value at time k, b is the bias value, F is the activation function, y(k ) represents the output value at time Wi (k ). The activation function between the hidden layer and the output layer is:
步骤2.3通过上式得到语义特征y(k)后，接下来用余弦相似度公式计算用户文本和问题文本之间的相似性，公式如下：Step 2.3 After obtaining the semantic feature y(k) through the above formula, then use the cosine similarity formula to calculate the similarity between the user text and the question text, the formula is as follows:
其中，yQ,yU分别表示问题文本和用户文本的概念向量。Among them, yQ , yU represent the concept vector of question text and user text, respectively.
步骤2.4用归一化指数函数(softmax)计算后验概率。用归一化指数函数，将计算社区中新提出的问题文本和社区中已经存在的用户文本之间的语义相似性得分转化为一个求后验概率的问题，公式如下：Step 2.4 Calculate the posterior probability with a normalized exponential function (softmax). Using the normalized exponential function, the calculation of the semantic similarity score between the newly proposed question text in the community and the existing user text in the community is transformed into a problem of finding the posterior probability. The formula is as follows:
其中，U表示用户文本，Q表示问题文本，N是样本空间中U的数量，yQ表示输出层问题文本向量，yU表示输出层用户文本向量，表示第n个用户输出层的用户文本向量。where U represents user text, Q represents question text, N is the number of U in the sample space, yQ represents the output layer question text vector, yU represents the output layer user text vector, Represents the user text vector of the nth user output layer.
步骤3，针对候选专家群构建用户问答关系有向图，从所有提出问题的用户出发建立链接，链接方向指向为该问题提供答案的用户。以图3所示的6个用户为例，用户u1提出了问题q1和q3，用户u2提供了答案a1，用户u4提供了答案a3，这样用户u1和用户u2之间建立一个链接，由u1指向u2；用户u1和用户u4之间建立一个链接，由u1指向u4。用户u2提出了问题q2和q4，用户u3提供了答案a2，用户u4提供了答案a4，这样用户u2和u3之间建立了一个链接，由u2指向u3；用户u2和u4之间建立了一个链接，由u2指向u4。用户u3提出了问题q6，用户u5提供了答案a6，这样用户u3和u5之间建立了一个连接，由u3指向u5。用户u4提出了问题q5，用户u6提供了答案a5，这样用户u4和u6之间建立了一个链接，由u4指向u6。过滤掉问题和答案之间的关系，直接转化成只包含用户节点的用户问答关系有向图。Step 3: Constructing a directed graph of user question-answer relationships for the candidate expert group, establishing links from all users who asked questions, and the link direction points to users who provide answers to the question. Taking the 6 users shown in Figure 3 as an example, user u1 asked questions q1 and q3 , user u2 provided the answer a1 , and user u4 provided the answer a3 , so that user u1 and user u2 A link is established between users, and u1 points to u2 ; a link is established between user u1 and user u4 , and u1 points to u4 . User u2 asked questions q2 and q4 , user u3 provided the answer a2 , and user u4 provided the answer a4 , so that a link was established between users u2 and u3 , from u2 to u3 ; A link is established between users u2 and u4 , and u2 points to u4 . Useru3 asks question q6 , user u5 provides answer a6 , so that a connection is established between user u3 and u5 , from u3 to u5 . User u4 asks question q5 and user u6 provides answer a5 , so that a link is established between user u4 and u6 , from u4 to u6 . Filter out the relationship between questions and answers, and directly convert it into a directed graph of user question-answer relationships that only contains user nodes.
步骤4，使用主题敏感性回答者排序算法(TSAR)对用户问答关系有向图进行链接分析，按照如下步骤4.1至4.2计算各用户的权威度值，选出权威度值最高的TOP-N个结果，即选出权威度值最高的N个用户作为最终推荐的专家列表。Step 4: Use the Topic Sensitivity Responder Ranking Algorithm (TSAR) to perform link analysis on the directed graph of user question and answer relationships, calculate the authority value of each user according to the following steps 4.1 to 4.2, and select the TOP-N with the highest authority value. As a result, the N users with the highest authority value are selected as the final recommended expert list.
步骤4.1将用户问答问题的数量、问题的难度值和回答问题的平均响应时间量化为随机转移概率矩阵的权重，权重的计算公式为：Step 4.1 Quantify the number of questions asked by the user, the difficulty value of the question and the average response time of answering the question as the weight of the random transition probability matrix. The calculation formula of the weight is:
其中，Nij表示用户uj累计回答ui问题的个数，表示用户提出的问题难度值，Tavg表示用户uj对ui提出的问题的平均响应时间。用户提出的问题难度值的公式为：Among them, Nij represents the cumulative number of user uj answers to ui questions, Represents the difficulty value of the question raised by the user, and Tavg represents the average response time of the user uj to the questions posed byui . The difficulty value of the question asked by the user The formula is:
其中，表示用户uj回答用户ui提出的问题的难度值，表示用户uk回答问题ui的时间，表示用户ui提出问题的时间，N表示ui提出的问题获得的答案总数。用log函数解决难度值的长尾分布问题。用户uj对ui提出的问题的平均响应时间Tavg的公式为：in, represents the difficulty value of user uj answering the question raised by userui , represents the time when useruk answers questionui , Represents the time when userui asks the question, and N represents the total number of answers to the questionui raised. Use the log function to solve the long-tailed distribution of difficulty values. The formula for the average response time Tavg of user uj to the questions posed byui is:
其中，Tjk表示用户uj回答用户ui的第k个问题的时刻，Tik表示用户ui提出第k个问题的时刻，Nij表示用户uj回答用户ui提出的问题总数。Among them, Tjk represents the moment when user uj answers the kth question of userui , Tik represents the moment when userui asks the kth question, and Nij represents the total number of questions that user uj answers to userui .
步骤4.2迭代计算用户的权威度，用户的权威度计算公式为：Step 4.2 Iteratively calculate the authority of the user. The calculation formula of the authority of the user is:
其中β是阻尼因子，N′表示构造的有向图中节点的个数，表示ui到uj的权重，∑ui表示ui到所有用户的权重和。where β is the damping factor, N′ represents the number of nodes in the constructed directed graph, Represents the weight of ui to uj , and ∑ui represents the weight sum of ui to all users.
本实施方式的方法将文本语义分析技术和改进的主题敏感性排序算法联合使用。首先使用经典的深度结构化语义模型(DSSM)对社区中的文本进行语义层面的分析，根据文本语义相似度值的大小生成候选专家群并构建用户问答关系有向图；然后使用改进的主题敏感性排序算法对有向图进行链接分析，该算法在构建随机概率转移矩阵的过程中充分考虑了用户回答问题的数量、问题的难度系数值和用户回答问题的平均响应时间，解决了已有方法推荐准确性低和专家提供回复不及时等问题，该方法推荐的专家不仅具有较高的兴趣度和专长度，而且还有一定的权威性。本方法可以应用于对知识问答社区中新提出的问题进行专家推荐。例如，在Yahoo！Answers知识问答社区中用户提交了一个自己困惑的并急于获取答案的问题，该方法可以为其推荐有能力有兴趣解决该问题并且可以及时回复的专家，使用本方法可以提高知识问答社区的运转效率，有效改善用户的问答体验。The method of this embodiment uses the text semantic analysis technology and the improved topic sensitivity ranking algorithm in combination. Firstly, the classic deep structured semantic model (DSSM) is used to analyze the text in the community at the semantic level, and the candidate expert group is generated according to the size of the text semantic similarity value and the user question and answer relation directed graph is constructed; then the improved topic-sensitive The linearity sorting algorithm performs link analysis on directed graphs. In the process of constructing a random probability transition matrix, the algorithm fully considers the number of questions answered by users, the value of the difficulty coefficient of questions and the average response time of questions answered by users, which solves the problem of existing methods. For problems such as low recommendation accuracy and untimely responses provided by experts, the experts recommended by this method not only have a high degree of interest and expertise, but also have a certain degree of authority. This method can be applied to expert recommendation for newly posed questions in the knowledge question answering community. For example, on Yahoo! A user in the Answers knowledge question and answer community submits a question that they are confused and eager to get an answer to. This method can recommend experts who are interested in solving the problem and can reply in time. Using this method can improve the operation efficiency of the knowledge question and answer community. , which can effectively improve the user's Q&A experience.
本发明实施例如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明实施例的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机、服务器、或者网络设备等)执行本发明各个实施例所述方法的全部或部分。而前述的存储介质包括:U盘、移动硬盘、只读存储器(ROM,Read Only Memory)、磁碟或者光盘等各种可以存储程序代码的介质。这样，本发明实例不限制于任何特定的硬件和软件结合。If the embodiments of the present invention are implemented in the form of software function modules and sold or used as independent products, they may also be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present invention may be embodied in the form of software products in essence or the parts that make contributions to the prior art. The computer software products are stored in a storage medium and include several instructions for A computer device (which may be a personal computer, a server, or a network device, etc.) is caused to execute all or part of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: U disk, removable hard disk, Read Only Memory (ROM, Read Only Memory), magnetic disk or optical disk and other media that can store program codes. As such, embodiments of the present invention are not limited to any particular combination of hardware and software.
相应的，本发明的实施方式还提供了一种计算机存储介质，其上存储有计算机程序。当所述计算机程序由处理器执行时，可以实现前述用于网站知识社区系统的专家推荐方法。例如，该计算机存储介质为计算机可读存储介质。Correspondingly, embodiments of the present invention also provide a computer storage medium on which a computer program is stored. When the computer program is executed by the processor, the aforementioned expert recommendation method for the website knowledge community system can be implemented. For example, the computer storage medium is a computer-readable storage medium.
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 flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows 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 flows 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.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910598556.8ACN110321421B (en) | 2019-07-04 | 2019-07-04 | Expert recommendation method for website knowledge community system and computer storage medium |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910598556.8ACN110321421B (en) | 2019-07-04 | 2019-07-04 | Expert recommendation method for website knowledge community system and computer storage medium |
| Publication Number | Publication Date |
|---|---|
| CN110321421Atrue CN110321421A (en) | 2019-10-11 |
| CN110321421B CN110321421B (en) | 2023-06-20 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910598556.8AActiveCN110321421B (en) | 2019-07-04 | 2019-07-04 | Expert recommendation method for website knowledge community system and computer storage medium |
| Country | Link |
|---|---|
| CN (1) | CN110321421B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111667200A (en)* | 2020-07-09 | 2020-09-15 | 腾讯科技（深圳）有限公司 | Authority determination method, device, equipment and storage medium |
| CN112434517A (en)* | 2020-11-09 | 2021-03-02 | 西安交通大学 | Community question-answer website answer ranking method and system combined with active learning |
| CN112765326A (en)* | 2021-01-27 | 2021-05-07 | 西安电子科技大学 | Question-answering community expert recommendation method, system and application |
| CN112800208A (en)* | 2021-01-21 | 2021-05-14 | 中国人民解放军国防科技大学 | Question reviewer automatic recommendation method and system for question answering community |
| CN114329184A (en)* | 2021-12-03 | 2022-04-12 | 南京邮电大学 | A method and system for expert recommendation in a question and answer website |
| CN117763240A (en)* | 2023-11-27 | 2024-03-26 | 合肥工业大学 | Community expert recommendation method and system considering personality traits and theme preferences of users |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150032492A1 (en)* | 2013-07-24 | 2015-01-29 | Lithium Technologies, Inc. | Methods of Identifying Relevant Content and Subject Matter Expertise for Online Communities |
| 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 |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150032492A1 (en)* | 2013-07-24 | 2015-01-29 | Lithium Technologies, Inc. | Methods of Identifying Relevant Content and Subject Matter Expertise for Online Communities |
| 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 |
| Title |
|---|
| ERIKS SNEIDERS 等: "Automated FAQ Answering with", 《2009 2ND CONFERENCE ON HUMAN SYSTEM INTERACTIONS》* |
| 周蓓蓓;明晓乐;孙婷婷;: "知识社区中用户问题分配方法与实现"* |
| 颜昕 等: "基于自然语言处理的医疗健康问答系统", 《通讯世界》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111667200A (en)* | 2020-07-09 | 2020-09-15 | 腾讯科技（深圳）有限公司 | Authority determination method, device, equipment and storage medium |
| CN111667200B (en)* | 2020-07-09 | 2023-09-19 | 腾讯科技（深圳）有限公司 | Authority determination method, device, equipment and storage medium |
| CN112434517A (en)* | 2020-11-09 | 2021-03-02 | 西安交通大学 | Community question-answer website answer ranking method and system combined with active learning |
| WO2022095573A1 (en)* | 2020-11-09 | 2022-05-12 | 西安交通大学 | Community question answering website answer sorting method and system combined with active learning |
| CN112434517B (en)* | 2020-11-09 | 2023-08-04 | 西安交通大学 | A method and system for sorting answers of community Q&A websites combined with active learning |
| US11874862B2 (en) | 2020-11-09 | 2024-01-16 | Xi'an Jiaotong University | Community question-answer website answer sorting method and system combined with active learning |
| CN112800208A (en)* | 2021-01-21 | 2021-05-14 | 中国人民解放军国防科技大学 | Question reviewer automatic recommendation method and system for question answering community |
| CN112765326A (en)* | 2021-01-27 | 2021-05-07 | 西安电子科技大学 | Question-answering community expert recommendation method, system and application |
| CN114329184A (en)* | 2021-12-03 | 2022-04-12 | 南京邮电大学 | A method and system for expert recommendation in a question and answer website |
| CN117763240A (en)* | 2023-11-27 | 2024-03-26 | 合肥工业大学 | Community expert recommendation method and system considering personality traits and theme preferences of users |
| Publication number | Publication date |
|---|---|
| CN110321421B (en) | 2023-06-20 |
| Publication | Publication Date | Title |
|---|---|---|
| US20240414109A1 (en) | Systems and processes for operating and training a text-based chatbot | |
| CN110321421B (en) | Expert recommendation method for website knowledge community system and computer storage medium | |
| US9990609B2 (en) | Evaluating service providers using a social network | |
| CN112667794A (en) | Intelligent question-answer matching method and system based on twin network BERT model | |
| CN110175227A (en) | A kind of dialogue auxiliary system based on form a team study and level reasoning | |
| Wang et al. | Student physical health information management model under big data environment | |
| US11636411B2 (en) | Apparatus for determining role fitness while eliminating unwanted bias | |
| CN115017887B (en) | Chinese rumor detection method based on graph convolution | |
| CN110111010B (en) | A method and system for question answering task assignment based on crowd intelligence network | |
| CN116775882B (en) | Intelligent government affair message processing method and equipment | |
| KR102624636B1 (en) | Self-feedback based questionnaire creation system | |
| CN115191002A (en) | Matching System, Matching Method, and Matching Procedure | |
| WO2023279631A1 (en) | Speech manuscript evaluation method and device | |
| CN119316389A (en) | A dialogue method, device, equipment and medium | |
| CN118861438A (en) | A digital consulting management method and system for engineering information based on the Internet of Things | |
| CN117217808B (en) | Intelligent analysis and prediction method for activity invitation capability | |
| CN114154839A (en) | Course recommendation method based on online education platform data | |
| CN108959467A (en) | A kind of calculation method of question sentence and the Answer Sentence degree of correlation based on intensified learning | |
| CN117312676B (en) | A smart reading recommendation and collaboration analysis method | |
| CN118656659A (en) | Job matching method, device, computer equipment, and storage medium | |
| Khayrullin et al. | Predicting psychology attributes of a social network user | |
| Liu et al. | Finding social media trolls: Dynamic keyword selection methods for rapidly-evolving online debates | |
| Lei | Research on personalized education recommendation algorithm based on artificial intelligence | |
| Verma et al. | University chatbot system using nlp | |
| CN115511181A (en) | A cross-domain rumor propagation prediction method based on transfer learning |
| 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 |