From Wrapping to Knowledge Domain Ontology Learning from Dee

Zhiming Cui 1 , Pengpeng Zhao1 , Wei Fang1,2 , Chao Lin1 1 The Institute of Intelligent Information Processing and Application2 Jiangsu Key Laboratory of Computer Information Processing TechnologySoochow University,Suzhou, 215006, Chinaszzmcui@AbstractThe next generation of the Web, called SemanticWeb, has to improve the Web with semantic pageannotations to enable knowledge-level querying andsearches. However, manual construction of theseontologies is a time consuming and difficult task. In thispaper, we describe an automatic extraction method thatlearns domain ontologies for semantic web from deepweb. Our approach first learns a base ontology from deepweb query interfaces, then grows the current ontology byprobing the sources and discovering additional conceptsand instances from the result pages. We have evaluatedour approach in several real-world domains. Preliminaryresults indicate that the proposed extraction methoddiscovers concepts and instances with high accuracy.1. IntroductionNowadays the Internet contains a huge collection ofdata stored in billions of pages and it is used for theworldwide exchange of information. The pages mainlyrepresent textual data and have no semantic annotation.Thus, query processing based mostly on inefficientkeyword matching techniques becomes a bottleneck ofthe Web. Tim Berners-Lee coined the vision of the nextversion of the Web, called Semantic Web [1], in which