Research On Lightweight Personalized Search Engine System

Using Web Search Engine has become an important way for users to get information from the Internet. Users hope that the Search Engine can provide all-around information resources. How to get the most valuable information through an effective way for users is becoming more and more necessary, which is also becoming an indispensable service that the Personalized Search Engine have to provide.Users' expectation of Search Engine is to find a way for customizing the engine according to their own interests, which is an issue of how a search engine can provide personalized service in the realm of information search. Although the traditional server-based Search Engine can meet users' demands of personalized search service, it causes the problem of users' behavior database overload in that users should frequently communicate with the database on the server side.Based on what has been discussed above, this thesis provided a new style search structure, a Lightweight Personalized Search Engine, to achieve the aim of reducing the servers' load. This kind of search structure can also solve the problem of the sort error caused by client-sorted users' interests. This kind of structure combines the client storage resources and server-side users' interests database, and the client-side agent technology is introduced to solve the problem of the servers' overload. In addition, the loose-coupled connection of the client, the client-agent and the server can improve the agility of personalized search engine. This thesis emphasizes the communication rule of the client, the client-agent and the server and the issue of how to reduce the communication traffic among them is discussed in detail. The thesis gives the users' interests model updating algorithm on the client, the client-agent and the server. Finally, an experiment is used to prove the feasibility of the Lightweight Personalized Search Engine. Through the Comparative Analysis between the Lightweight Personalized Search Engine and the traditional Search Engine, this kind of search structure can really help improve the server's performance to a certain extent.