Research Of BBS User Evaluation System Based On Data Mining

With the growing development and popularity of the network, information is being produced explosively and disseminated on the network. BBS gradually becomes a platform for communicating and expressing ideas. How to use the massive information on the network to rate and classify users of BBS has become an important proposition which would help administrators manage the BBS and make it better. In this paper, the recently rising poem forums are selected to be research object. The main content of paper is as follows:Based on web page information collection and extraction, propose a method to remove noise text by calculating relation degree of text using a triple word matching algorithms.Polarity dictionary construction. The first that used to general emotion analysis is built by selecting the words of HowNet artificially. After the word segmentation of over 100 poem comments, the second dictionary is constructed by word similarity calculation, which is used for polarity calculating in profession field.Analysis of topic hotness and text polarity. In this paper, the polarity of text is being calculated differently separated by classical and vernacular. For the first time, propose a method to analysis classical text polarity.Based on analysis of topic hotness and polarity, average text length, average support rate, number of participation topics, topic number and reply number are selected as factors to rate and classify users of BBS by using Decision Tree method and then generated rules.