Application Of Knowledge Base System In Meta-search Engine

Traditional search engine (SE) works with the full-text model, it is simple, fast,distributed, and has now become the Internet’s most important information retrieval tool.But traditional SE can not understand the user’s real intentions and the semanticinformation of the date on the Web. As a result, it will fail when user tries to retrievalinformation according to the correlation among concepts and leads to the poor recallratio and the precision of retrieval results. In addition, traditional SE is unable to realizeknowledge retrieval.Its well knows that semantic web is the best method to resolve above problems.Ontology is an effective way to construct semantic web. It has characteristics ofconceptualization, explicit, formal and share. By ontology, knowledge base (KB) whichsupports semantic retrieval and reasoning can be constructed.Meta-search engine can takes use of the advantages of member SE, and fuse theresults of a query, so it can obtain optimal effects. Because meta-search engine don’tneed to create its own index database, so this paper use it as a platform of ontologyknowledge base retrieval.In order to further study of ontology retrieval, this paper presents an intelligentsearch engine framework based on ontology. This framework can help SE change itsworking model from full-text retrieval to knowledge retrieval, and improve the recallratio and precision of SE. Ontology KB can be used to extend semantic retrievalfunction of meta-search engine, and make it possible to retrieval web and semantic webat the same time, and give references between each other. This mechanism helps usersto make accurate judgment about semantic retrieval results.All aspects of this framework involve are discussed in detail in this paper,including the construction of OWL KB, ontology operation based on jena, ontologyinference based on SWRL, and analyzes the function and working mode of eachsubsystem in the framework. This system retrieval ontology KB by using intelligentretrieval agent so as to reduce the users’ workload, and enhance the search engine theability of dealing with complex inquires.Meanwhile, for heterogeneous and distributed ontology on the Internet, a discoveryand invoking model based on Web Service is presented. Web Service is a very matureInternet software infrastructure, which is supported widely by standard protocols.Extending the tModel in UDDI, Web Service can describe its profile in the way of semantic through OWL-S language.Finally, this paper introduces practical retrieval cases, which is respectivelyprocessed in current SE and the meta-search engine based ontology mentioned above.With the comparison and analysis of the different results, it can be concluded that thelatter can get more satisfactory value of recall ratio and precision and greatly reduceuser’s retrieval steps and retrieval time.Although the above example is implemented in the ideal of experimentalenvironment, the combination of SE and ontology can motivate the development of SEindeed.