| In recent years,the development of online question answering system is more and more rapid,more and more users seek for information through online question answering system.Such as Baidu Zhidao,Quora,Yahoo! Answers,these online question answering systems has become the important way for users to obtain valuable information and share knowledge information.Through the online question answering system,the user can ask questions,answer the questions and evaluate the existing question answering information.However,with the development of online question answering system,more and more question and answer information is in the question answering system.A user has just submitted to the system a new problem,the new problem is submerged by new questions.In this case,it may cause users to wait for a long time to get answers.In addition,many of the answers contain a lot of spam or advertisement and other worthless information.Therefore,it is necessary to study a kind of expert discovery method,so as to quickly find the user who can answer the question,and improve the efficiency.The main work of this paper is as follows:1.Summarize the existing research results of expert discovery in online question answering system,including the principle of PageRank algorithm and LDA topic model which need to be used in the process of expert discovery.Through the analysis of the traditional PageRank algorithm,we find the inherent defects of the algorithm,and improve it can better find the expert users in the online question answering system.2.Propose a method based on exponential smoothing.The idea of this algorithm is to add the time factor to the traditional PageRank algorithm.In the process of expert discovery,we use the exponential smoothing method to predict the number of users posted in Stack overflow.At the same time,we also add the trend factor to predict the future of of the user posting.Finally,the effectiveness of the improved algorithm is proved by the experiments in this paper.3.Propose an expert discovery method based on Bayesian prediction.The idea of this algorithm is not only to improve the traditional PageRank algorithm from the time,but also to improve the PageRank algorithm from the topic semantics.In the process of expert discovery,we use the Bayesian forecasting method to predict the user's post in Stack overflow.Finally,the experiment shows that the improved PageRank algorithm is better than the traditional PageRank algorithm in the P@n_Percent evaluation results. |
