The Design And Implementation Of Recommendation Strategy For Agent-based Intelligent Meta Search Engine

As the Internet has entered the age of "Web2.0", the user is not only a role that browses information but also a role that at the same time makes new information. This causes the dramatically incensement of the network information. Although the information retrieval technology also has a great development, the traditional search engine still fails to follow the increasing user information demand. However, meta search engine technology can overcome these shortcomings of the traditional search engine. It can provide a better search coverage rate, and the most important thing is that it can actively understand user’s instruction instead of only waiting for the orders from users like a traditional engine. Recommender system can use thus user information in order to provide an active information service to users. Therefore a combination between the recommender system and the intelligent meta search can maximize this intelligence, thus providing a better search experience. However, this combination between meta search engine and the recommender system still exist many problems in both research and application field and is worth an in-depth study.This paper focuses on the design,implement and application of recommender system which is based on an Agent-based intelligent meta search engine platform. First of all, this paper introduces the related theory of traditional search engines and recommender system as well the Agent-related theory and technology. Then gives the structure design of this Agent based search engine platform, design goal and working process of each module assembly. Moreover, using this platform as the foundation, this paper introduces two concepts which are data flow diagram and the user intention recognition, and puts forward a new recommendation mechanism design. This mechanism can not only adapt to the chinese environment, but also can make full use of the characteristics of meta search platform environment. Then, the paper gives the realization of the claimed recommender system and points out the critical points for combining it with the meta search platform.Finally, in order to verify the validity and rationality of the proposed recommender system which is rely on the Agent based meta search engine platform, this paper designsa experimental demonstration which consist of two different stages. First of all, through the off-line testing means which does not depend on the particular meta search engine platform, the paper quickly tests recommender system’s performance and recommendation’s rationality. Then through the online experiment, it again confirms that the recommendation mechanism proposed in this paper can also make good recommendations in the actual application situation, which can both take care of the relevance of the recommendation and the diversity of the recommendation.