Research On A Multi-ontologies Aided Semantic Annotation Model

The gigantic volume of web content forces people to rely on machines to help search for information. Nowadays, the search engine can be used to retrieve information but is not enough, because the machine itself can not understand the content of web. The Semantic Web is a straightforward solution to this problem. The Semantic Web is an extension of the current web in which information has well-defined meaning, is understandable by computers and is convenient for the interactivities between people and machine. Semantic annotation is to tag Instances of ontology class and map them into ontology classes. The realization of the Semantic Web requires the widespread availability of semantic annotations for existing and new documents on the Web.Recently, many Semantic Annotation platforms are supported by a single ontology, but Web pages often cover more than one field. Therefore, the paper proposes one Multi-ontologies Aided Semantic Annotation Model (MASAM), which includes four main modules:ontology integration, information extraction, information annotation and information retrieval. Ontology integration module firstly integrates relevant ontologies then converts integrated ontology knowledge to JAPE rules. Information extraction module extracts concepts, instances and relations from Web pages with the help of JAPE rules. Information annotation module gives a new algorithm to calculate the correlation degrees among documents, and stores the annotation information into database, separating from Web pages. Users transfer Information retrieval module to get the needed information from the annotation repository.This thesis also designs and implements a prototype system (Multi-ontologies Aided Semantic Annotation System, MASAS), which is provided to users through Semantic Web Services. The experimental results reveal that the framework and algorithm are feasible and effective.