| Summary: | 碩士 === 朝陽科技大學 === 資訊管理系碩士班 === 94 === Ontology can be used to build metadata which describes data about data and offers a group of glossaries with individual definition that covers a certain knowledge area. It not only transfers syntax of words but also accurately transfers semantic data between human users and the network. Hence, the usefulness of the semantic web depends on whether the domain of ontology can be constructed effectively and correctly. In this thesis we propose an automated method to construct an domain ontology. First, we collected domain-related web pages from the Internet. Secondly, we use the HTML (Hypertext Markup Language) tag labels to choose meaningful terms from the web pages. Next, we use these terms to construct an domain ontology by calculating a TF-IDF (Term Frequency-Inverse Document Frequency) to find the weight of terms, using a RART network (Recursive Adaptive Resonance Theory Network) to cluster terms. Each group of terms will find a candidate keyword for ontology construction. Boolean operations locate individual keywords in a hierarchy. Finally, the system outputs an ontology in a Jena package using an RDF (Resource Description Framework) format. The primary experiment indicates that our method is useful for an domain ontology creation.
|
|---|
