The process of development of an ontology-based knowledge portal and creation of its content is time-consuming and labor-intensive. It is very desirable to have special analysis facilities for maintenance and development of such a portal. The subject of our paper is a tool for visual analysis of content and ontology of a knowledge portal.

The basis of our approach is application of graph generation and graph visualization methods. Browsing of the content of a knowledge portal is organized as a multilevel stepwise process. Appropriate placement algorithms are used at each step of this process. They take into account certain types of ontological relations and their combinations. Radial, circular and layered tree placement algorithms are used for visualization of the inheritance relation. Several force-directed algorithms are used for visualization of role relations between different classes as well as between class instances. Finally, we have two special placement algorithms for visualization of the superposition of partonomy and role relations. A number of useful graph transformations are implemented. They essentially improve understandability of a portal under investigation.

This approach provides users with general understanding of ontology and a possibility of visual estimation of the sizes of classes and relations. It allows the user to select and analyze specific relations between classes and between objects as well as to identify errors. The program was tested on real knowledge portals and appeared to be rather helpful.

apanovih.pdf880.43 KB