The Research On Getting The Search Intent Of Users

With the development of the Internet, people have been used to get information from it. The information in Internet is poor organized and its amount is very huge. It has become an urgent problem to get useful information from Internet effectively. The appearance of the search engine solved this problem at a certain extent. The characters of keyword matching retrieval model, which is adopted by almost all general search engines, are somewhat mechanical and it is difficult to overcome the phenomenon of synonyms in natural language. These make a lot of useless results and the retrieval efficiency is correspondingly low. In our research we hope to get the intent of the user when he or she submits a certain keyword, and the search results are optimized with this user intent.We found the category attribute of words when we analyze the web pages which contain a certain word with the web page categorization technology. Most words make sense in one category or some categories. So we use the category attributes of words to represent the user intent. According to the difference of the category attribute of words, words can be classified to"single-meaning words"and"multi-meaning words". Correspondingly, the distilling of the user intent can be classified to"the distilling of the common user intent"and"the distilling of the personalized user intent".The"single-meaning word"oriented distilling is called"the distilling of the common user intent". Because a"single-meaning word"only makes sense in one category, so the intent of different users who submit a same"single-meaning word"can be considered to be same. And the problem of the distilling of the common user intent can be transformed to getting the category attribute of words. We introduce a latent semantic analysis base method of getting the category attribute of words in this paper, which has been proved to be effective by experiment.The"multi-meaning word"oriented distilling is called"the distilling of the personalized user intent". Because a"multi-meaning word"makes sense in several categories, so the intent of different users who submit a same"multi-meaning word"may be different. Consequently, the personal interest of users will be needed to make further judgment. We analyzed the web pages which are browsed by users to research their interest. The analysis result of one user is listed in this paper. At last, we introduce the method of the distilling of personalized user intent according to these analyses.