The Research On User Profile Of Intelligent Meta-Search Engine

In recent years, meta-search engine has been applied widely because it has thetrait of basing on underlying search engines and can offer more retrieval help to users.But, in the movement, it is important more and more that user could acquire perfectindividuation services provided by meta-search engine. Ulteriorly, it is expected thatcryptically retrieval requests can be offered effective services more to user frommeta-search engine to help he makes clear the true retrieval intentions.Aiming at the two problems, at first, the paper summarizes research productionson a series of correlative contents of meta-search engine, user profile and reasoning.Then, it designs an intelligent meta-search engine model based on Agent (M.Meta-search Engine, MMSE) and researches several primary realization technologieson the bases of analyzing ecumenical model of meta-search engine. In succession, inallusion to user's demands of individuation and intelligence, the paper advances amodeling method of user profile expressed with two conception layers in theory, treegraphics and mode spatial graphics based on ontology, and studies an algorithm onfounding and updating of user profile. And then, to meet user's cryptically retrievalrequests, this paper brings forward a sort of reasoning mechanism based on the userprofile, advances reasoning algorithm and analyzes it theoretically. At last, the authordesigns a retrieval interaction interface prototype on classified hiberarchies of theMMSE to incarnate the important effect of above works to information retrieval.The innovative attempts of the paper are listed as follows:1) It designs an intelligent meta-search engine model based on Agent (MMSE)on the bases of analyzing ecumenical model of meta-search engine at length whichhas applied value at a certain extent. This paper studies a sort of new schedulerstrategy of underlying search engine based on search engine repository, and thestrategy chooses right engines according as important information of every searchengine that is collected and deposited in structures such as weight vectors matrix based on character lemma. On the bases of probabilistic models, it studies a sort ofcompositor method of classified correlation. The benefit of classified idea is not onlyto let it faster that user finding needful information, but also to let it more convenientthat the MMSE collecting user's feedback information. Hereby, the paper designs theretrieval interaction interface prototype on classified hiberarchies of the MMSE.2) It advances a modeling method of user profile expressed with two conceptionlayers in theory, tree graphics and mode spatial graphics based on ontology. Theinitialization value of the tree graphics is based on Dewey decimal classification afterbeing perfected. It makes the MMSE having an effect of concentrating on the mainpoints on the whole when looking for right mode spatial graphics from the treegraphics. The paper studies the mode spatial graphics based on ontology at length,sets up ontology nodes in spatial graphics and introduces the theory of interval valuedfuzzy sets. In addition, it explains the method in theory, brings forward ontologyrepresentation of nodes and a series of correlative definitions and formulae, andfounds a corresponding mathematical model. The modeling method is helpful toameliorating deficiencies of traditional modeling methods and is in favor of userprofile's foundation, use and perfection more. Finally, it puts forward an algorithm onfounding and updating of user profile and designs a series of correlative datastructures and subject functions. The paper uses an implement of well-ordered toprove correctness of the algorithm strictly, and then, analyzes time complexity of thealgorithm which is T(n)=o(n~2). On the theory, the algorithm is proved of the traits ofcorrectness, validity and low time complexity. It should be considered that theauthor's works is a useful attempt on the research of user profile.3) It brings forward a sort of reasoning mechanism based on the user profile anda reasoning algorithm of the user profile. The whole of user profile makes up of anaction background when user searching information. And the background idea isfeasibility to reason user's real retrieval requests according to analyzing from pointsof psychology and artificial intelligence. The reasoning mechanism is advanced tomeet user's especial retrieval demands, for instance, user can not expressly describehis real demand. On the theory, the algorithm is proved of the traits of correctness,validity and low time complexity in the same analysis method.The author's works is sponsored by the Grand Science and Technology ResearchProgram of Ministry of Education, the Research on Individuation Technology of Web Information Retrieval based on Agent (03020), and the two Natural ScienceFoundation of Shanxi Province, the Research on Technology of Individuation WebInformation Retrieval and Agent (20031038) and the Research on OntologyDescription Method of Image Semantic Sensibility. Therefore, the paper chooses it asresearch content of individuation and intelligence of meta-search engine. The target ofit is to provide pertinence service to each user aiming at his unique informationdemands, thereby, to meet his retrieval requests of individuation and intelligence. Theworks of this paper provides and perfects a sort of sound information service mannerand has important significance in theory and applied value to search information byfully utilizing resource in Internet, to enhance use efficiency of information resourceand efficiency of retrieval, and to implement people'individuation requirements in theinformation times. Hence, the author's works is advanced and innovative in a certainextent for its research content is developmental direction and hotspot of informationretrieval.