Studying And Design Of Semantic Search Engine Technology Based On Personalized Ontology

With the rapid development of the network, the online information is increasing quickly. The mass of information is not only bringing people rich resources, but also bringing people great challenge on how to use the information effectively. The traditional search engine alleviated this problem to some extent. However, because it can not understand the semantics of the information, the quality and effectiveness of search is far to satisfy. The recall ratio and the precision ratio are to be improved. The search strategy of semantic search engine improves from traditional keyword level up to semantic level, with intelligent reasoning. This improves the performance of search. Semantic search technology achieves the semantic representation of information, knowledge reasoning and semantic relevance calculation. This technology can solve the current difficulties of the search techniques effectively.This paper introduces the working principle of search engine, classification and evaluation criteria at first. Then the paper researches the current development of the search engine in detail. And after that, it introduces relevant terms, researches the crucial technology, the principle, the development and the current situation of application, which relates to semantic search engines.Traditional search engine cannot understand the semantics. So this paper presents a semantic search engine based on personalized ontology, and combining with the technology and advantages of traditional search engine, designs a search engine Sendu which can increase the function of semantic understanding. Sendu engine uses the Heritrix as crawler in the network to retrieve information resources, and use Lucene framework to build traditional search engines as basis, in order to increase the semantic understanding ability, using ontology editor tool protégéto create domain ontology. Sendu engine can convert the user's query keywords into recognizable semantic terms by using ontology. Sendu uses RDF API - the jena for reasoning, and gets vocabulary which has semantic relation with users'search keywords, such as synonyms, upper word, etc. These extended terms will improve the recall ratio of search engine. For example, searching for "computer" this word, jena will deduce its synonyms of "PC"according to ontology knowledge base. So that final query will be expanded relevant web pages of the terms "computer" and "PC". Ontology will organize the term in a flexible way. Sendu engine also proposes three more semantic searches filter to filter out some web pages which users don't want to get. The prototype system implementation allows the user to take custom ways. For example, if the user does not consider PDF resources online, the semantic search filter will filter the documents in this format to solve the current search engines ignoring document resources of some formats.At last, this paper implements the prototype system of Sendu engine, which gains pretty good recall precision and recall ratio. We make further comments to improve the system on the base of the performance evaluation of the system. And the application of semantic search engine is prospected.