The Distribution Method And Implementation Of Users’ Questions For Knowledge Community

With the development of Web2.0, Knowledge Community Q & A system emerged as an important medium for users access to get information and knowledge sharing, and provides users with a convenient interactive platform. Such as Baidu, Yahoo!Answers and other Q & A community will release tens of thousands of questions every day, which makes the community Q & A system must have huge data repository and the ability to make a timely response to new questions submitted by users.However, with the rapid development of Knowledge Community, the questions raised by users in the community have a wide variety of types and huge data resources, leading experts need to spend more time looking for the problem of interest, so that users need to wait longer to get the answer. Meanwhile, with the number of candidate answers increasing, the users may have difficulty in choosing the best one from those with uneven quality. Finally, that result users can’t get better and timelier help.In response to these problems, this paper systematically studied the allocation method of the users’ questions, including the following two aspects:(1)Based on deep semantic analysis of the users’ questions, this paper promotes an area experts discovery method based on topic-modeling. This paper adopts the PLSA model to present experts, preference distribution according to experts’ previous answering history, thus generating recommending question lists. Combined with the similarity between the questions expert has answered and the users’ questions, get candidate related fields experts set.(2)After studying the answers’ characteristics of experts, this paper promotes a best experts discovery method based on link analysis technology. This paper adopts the PageRank algorithm to comprehensive measure the authority of experts in related fields and combined with the similarity between the relevant experts answered questions and matching answer, ultimately, get the best set of experts, to achieve a reasonable distribution of users’ questions.In this paper, the data sets are collected from Yahoo!Answers, the experiment’ result shows that the PLSA technique can be very effective in mining interest and improving the accuracy of the question recommended. For obtaining the optimal set of QA expert sets, using four link analysis methods to verify the experiment and two widely-used metrics MAP and AP@10 to evaluate the performance. The experiment results show that the additional weight of the PageRank algorithm can better measure the authority of experts.