This paper introduces the method of ontology based text classification. The class of ontology is associated with the related words and their weights from the pre-classified documents as a learning stage. In the main process, the words and their mutual relations are extracted from the target documents. Then the target document is mapped with classes of ontology. We experiment with International Patent Classification (IPC) ontology and huge pre-classified patent documents to classify scientific abstract.
The current goal of our research is to build a Japanese medical ontology. It becomes a key to the successful development of various knowledge processing applications for systematization of vast medical knowledge and the synthetic understanding of them. This article discusses a fundamental consideration of medical ontology based on ontological theory. We focus on "anatomical structure of organs" and "abnormal state in the human body". On the basis of the investigation, we distinguish organ-specific concepts from those independent of any organ. And we described abnormal states of a human body by the same framework which is represented by an object, its attribute and its value(abnormal type). These discussions give a guideline to fill the gap between different medical domains through establishment of common concepts across various branches of medicine.
While we have studied functional ontology and functional modeling framework for years, we developed an information system that enables users to describe function decomposition tree, which represents functional structure of an artifact, and successfully delivered it into industry. Recently, we renewed that system, which is called OntoGear, to adapt our new theory of ontology engineering with information contents centric architecture. OntoGear allows us not only to describe basic functional decomposition tree, but also to reveal a specific functional structure such as failure one. This paper discusses modeling of failure functional structure, and utilizing of it with using OntoGear.