The Research On The User Oriented Information Search System Based On Ontology

The full-text search engine becomes an important tool while using Internet. The technology used by current search engine is from the information retrieval field directly, and has been changed cosmically according to the characteristic of Internet. But because there are numerous fields and user level, the technology of the existing search engine can't meet users' demand more and more, with the rapid increase of Internet's data quantity. The proportion of the outlying links in the result increase rapidly. So it is very difficult for users to find the goal webpage smoothly with search engine.The present search engine not only returns the results briefly from the database, but also sort these results based on the relative grade between the result and user's intention. This is a principle of dependence known as "first law" of search engine. The effect of the algorithm influences the efficiency of a search engine. But the unified result returned by the common search engine is unable to meet the need of different users.The full-text search engine is based on the keywords matching technology. But there are so many multivocal, near-synonym, synonyms in the natural language, and the keyword that users offer might not exist in the purpose webpage. So the search engine can't reach the recall rate and precision required. Different from English, there is no natural participle marks in Chinese, which reduces the recall rate and precision further.In order to solve above two problems better, the concept search has received a large amount of recognition. But before the great breakthrough has not taken place in the artificial intelligence field, the concept search cannot get very good result. With rapid speed and high accuracy, vector space model become a standard technology for sorting at present.To solve the above two questions, this paper has proposed a kind of solution based on Ontology, have designed and realized the system of intellectual information search system of user oriented. The foundation of this method is that there might not be the keyword that users appointed in users' goal webpage, and the goal web pages can be got from other keywords. This search system forms the static ontology from the historical webpage, individual keywords database ofrelation. Then the search system translates the user keywords to dynamic ontology, system keywords required by search engine, and invokes the meta search engine, and then sorts the results according to relation, as users request. Thus on the basis of improving and checking the recall rate, the search engine can meet the demand for search efficiency.The evaluation of search engine is the recall rate and precision, which is derived directly from information retrieval field. Because of great data quantity, industry institute has already.no longer recognized these two evaluation criteria. These two concepts have become qualitative.Based on a method from TREC (Text Retrieval Conference) hold by NIST (National Institute of Standards and Technology), theme webpage search method, this search system has been evaluated, and the result has been analyzed. Experimental result indicates that this search system raises efficiency of theme webpage search.