Query Expansion Research In Personalized Intelligent Search Engine

With the continuous development of Internet, the amount of the networkinformation is increasing; people have become increasingly demanding on therecall rate, precision and personalized aspects of the search engine. Queryexpansion is a key link in the personalized intelligent search engine. Beforesearching the user’s query, Query expansion extends the user’s query effectively,which greatly improves the recall and the precision ratio of the search engine.Firstly, we extend the word entered by the user, and then use the TongyiciCilin and HowNet to calculate the similarity between the word user entered andthe word in Tongyici Cilin and HowNet. We find the words, which are the mostsimilar to the user input in Tongyici Cilin and HowNet, to extend the wordentered by user.Secondly, we expand the user’s query question. The realization of thisfunction consists of two parts. On the one hand, we extract and expand thekeywords of the query question. By eliminating the redundant part of thequery question and cutting the query question into words, we get the keywords ofthe query question. And then we expand these keywords as mentioned inparagraph above. On the other hand, we use commonly used words of questions’answers to expand the user’s query question. We classify user’s query question,and then collect the common used words of this type questions’ answers. At last,we use the common used words to expand the user’s query question.Then, we get user’s interests by analyzing user’s browsing behaviors.Through the analysis of the URLs in the user IE favorites and user browsinghistory, we extract the body of the corresponding web page; the body we getform web page was saved as a document. In this paper we use TF-IDF vectorspace model to generate a vector set, which corresponded to the set of documents.The vector set was clustered. Then we analyze the clustering results and extractthe words which can represent the user’s interests.Finally, query expansion and the user interest extraction were adding to thepersonalized intelligent search engine. First of all, we extend the query enteredby the user. Then, we retrieve the extended query in the search engine. Finally,the search results are sorted according to the user’s interest.