Analysis And Research Combining Based On The Structure Of The Network Information And Users’ Behaviors

With the increasingly diverse information needs, Community Question Answering (CQA) web services emerged. Because of extensive users’participation, an amount of information has been stored in them. These huge repositories of information provide a good content source for search engines. More and more users get information from CQA. However, the long-term accumulation of information causes a large number of "outdated" information, which is bringing inconvenience to users.Users’browsing logs can reflect their intent and habits. It is greatly significant for the analysis of CQA information timeliness. This paper proposed a method for intercepting query keyword from URL firstly. In addition, a method for normalizing query keyword, and a method for dividing query process for users’browsing logs are presented. Basing on these methods, a large number of real users’browsing historical records are statistically analyzed. The results show1) in average, a user browses eight web pages in six and a half minutes for one query;2) some queries are carried out concurrently and alternately;3) users rely on inner-site searching more.In this paper, a users’satisfaction judging framework for a CQA web page when it appears in the query-list is proposed based on the analysis of user browsing behaviors and CQA information inherent characteristics. Results show that the accuracy and recall rates of the classification have improved significantly by adding the characteristics of user browsing behaviors. By analyzing users’satisfaction and time-sensitivity of information in Yahoo Chiebukuro, it is found that users’satisfaction rate and time-sensitivity of information are different between different question categories.