The Research Of Domain-Oriented Search Engine Model Based On Meta-Search Scheduling

With development of internet and technique, the search engine is playing a more and more important role on its advantages; also it is concerned as a hot point recently. Because of mass, heterogeneity and dynamic character of information in internet, now the search engine has a great difficulty in meeting user's retrieval request when querying the web pages. Then, the meta-search emerges, it belongs to the kind of software system, forwards the user's query request to member search engine and deal with the result sent by every member search engine. Meta-search has improved the recall ratio in some degree, but it has difficulty in meeting user's accurate query request.In allusion to the problem of Information retrieval, in this paper, we propose a new domain-oriented search model based on meta-search. It improves the recall ratio through weighted domain-specific words module, this module is responsible for generating a domain-specific expression except the standard expression, then merging the two expressions as a whole input for the general search engine. This model adopts extended Boolean model for sorting the result set returned by the member search engines. The main research points in this paper are as follows:(1) We make a further research in Meta search scheduling strategy, analysis the advantage and disadvantage of these methods and the find the shortcomings, at last we proposed a new Meta search mix scheduling model based on learning. This model is considered from four perspectives: recall ratio, precision ratio, average response time and weighted user's interest. We considered every factor of influencing the domain-oriented search mode, then we analysis the algorithms.(2) Aiming at dissolving the restriction of search model in constructing their own index database, we construct a new domain-oriented search model based on Meta search, this model based on the general search engine and introduce the function module. The whole search model plays a great effect in improving query and meeting the user's request through cooperation of these several modules.(3) Constructing a domain keyword weighted module based on domain-oriented knowledge database. Though user's query information analysis and statistic, and also information reflection of register user, we can adjust the weight of domain keyword in order to update the domain-oriented keywords and the structure of the domain knowledge base dynamically.(4) We analysis the advantage and disadvantage of retrieval model,- then adopt sorting algorithms based on the extended Boolean Model for sorting the result sent by member search engines so as to improve user's satisfaction degree. At last, Design and make true of the realization of DOSM system.