An Ontology Learning Method Based on Double VSM and Fuzzy FCA

Ontology realization poses as the major hindrance to the evolution of World Wide Web into semantic Web. The manual construction of ontology demands large amount of labor as well as lasts for long durations. Ontology learning technology makes it possible for the automatic building of ontology in texts and greatly accelerates the speed of construction; yet, the technology is constrained by its lack of generality and accuracy. Based on the object-oriented analytical methods, a formal spatial description on ontology learning data source is performed. The authors revise the classical descriptive method for texts, that is, the object-oriented single vector space model (VSM) and propose a double vector space model (D-VSM), specifically, verbal layer and noun layer. This model is characterized by the inclusion of diverse attributes and solid relations. To further cope with information redundancy associated with FCA method and to improve the accuracy of concept abstraction, the fuzzy formal concept analysis (FFCA) ontology learning technology is introduced, which can fully explore the distributed property of data in DVSM and specialize in solving issues like ontology continuity, ontology relations obtainment, etc. An ontology learning tool is created based on DV-FFCA methodology, which provides powerful support for automatic (semi-automatic) ontology construction.