A KNOWLEDGE ACQUISITION MODEL FOR REPRESENTATION-INDEPENDENT KNOWLEDGE EXTRACTION FROM RDF

摘要

This paper presents a framework for knowledge extraction from Resource Description Framework (RDF) knowledge sources. The framework is "representation-independent" as it converts RDF-based knowledge into different representations, such as Prolog, relational database tables, and simple XML documents. Semantic Web has potential to improve the ability of disparate applications to acquire knowledge from RDF knowledge sources. However, different applications may require knowledge to be represented in different formats. Since each knowledge representation was designed for a specific purpose in mind, having multiple representations allows us to take full advantage of knowledge contained in RDF. For example, by converting RDF to Prolog, relationship among the RDF vocabularies can be correlated easily. By converting RDF to relational tables, statistics on the RDF structure can be drawn. By converting RDF to XML, knowledge in RDF can be re-represented as hierarchical structures. The different formats also improve a human's ability to comprehend the RDF structure and knowledge content. Our framework converts RDF into an intermediate internal knowledge representation, called the Universal Knowledge Format (UKF). UKF allows us to quickly and easily generate different target representations from RDF. This framework acts as a "knowledge middleware" to streamline knowledge extraction from RDF to other applications.