Analysis And Research On Personal Search Engine Based On User Interest

With the explosive development of internet and the increasing hugeness of World Wide Web(WWW), the complicated degree of web resources is growing by the exponential order quantity, causing the people to feel helpless when they search the information. In order to solve the problem of"the overloading information" and "the confusing resources", the search engine appears. Nowadays search engine often has its versatility, the arrangement mechanism has not considered to the user interest, thus it is very difficult to satisfy the user from different background, different intention. This article studies "the user interest profile"is for the user's sake of information resource on internet, it has the theory and the practical value.Firstly, the article describes the background of search engine, introduces the development of personal information service and its system structure, and sums up the current situation of search engine in the country and abroad. it introduces the basic theory about search engine dentally, summaries its definition and ingredient and their function, classifying search engine according to the different standard. The article gives the valuation standard of search engine, and from this way, it sums up the defect and developing trend of now available search engine.Secondly, the article introduces the key technology about building a profile in personal search, brings forward a kind of new method of user interest profile, that is to say, connecting the user handwork profile with the machine automation profile, leads into"relevance interest degree"according to user's browsing behavior. it expresses the degree of use interest to different web page, it is trained to renew the user interest profile by the nerve networks.Moreover, on the basis of analysis of tradition technology on result sorting,leading into the thought of facing subject and the premise calculating an impact of connection relation over page of page weight, it brings forward one kind of new the result ordering algorithm owing to interlink, and does a the experiment proving the feasibility.Finally, it designed a BP nerve network using the tool provided by Matlab to get user's relevance interest degree. it provides the process of its simulation, and analysis the result.