Research And Implementation On User-Oriented Query Expansion

With the fast development of Internet in recent years, Web resources increase at an explosive speed. By the beginning of 2004,the quantity of the webpage was up to 8 billion. At present the ways of using the resource on Web is various, the most popular is to use search engine, but the existing engines is for all users, which can't satisfy the user's individual demands, Personalized information service system can satisfy the users' individual demands effectively. Besides, some studies show that 58-81% webpage are those visited before[21], when people access web. So, Managing visited webpages effectively is meaningful for a personalized information service system.In most retrieval systems, the demand of users is represented by query keywords. In fact, there exists difference between the real demand of users and the query words. How to decreasing the difference is the key problem in implementing the user-oriented EIS. We put forward a way of query expansion oriented to user and adapted modification model oriented to users' interests. The model can increase the retrieval precision and make the returning webpages satisfy users better. Furthermore, using this model, we can ensure the numbers of query-words expanded and can optimize the regulating factor α,β,γ,λ .We cooperate to design an Electronic Information Assistant's prototype System based on individual, the idea of design is: first of all, when every registration of users, we request user provide his basic information, interest domain, query keyword and so on. We will create an initial user's interest model to every new user according to those basic information. Then we can search information by using existing search engine(such as Google, Baidu etc.). To the returning result, we filter the documents irrelevant to the user's interests and rerank the remaining documents according to user's profile. The remaining documents will be displayed in user interface. User can download and browse the documents which he is interest in, Personalized Electronic Information Assistant's System will modify the user's interest model and expand original query by the result of user feedback automatically, so that the interest model can really represent the users' interest. And we implement network information manage function, which can classify documents be searched EIS automatically.Further work: I. bettering the query expansion adapted model; 2. optimizing the regulating factor α,β,γ,λ ; S.perfecting the system's function.