Tools for Assembling and Managing Scalable Knowledge Bases

摘要

The DARPA High Performance Knowledge Base (HPKB) program was aimed to produce technology to rapidly construct large, reusable, and maintainable ontologies and knowledge bases (KBs). To achieve this goal, large-scale KBs cannot always be built from scratch, but instead need to be assembled as much as possible from existing resources. Reuse, however, does not come for free: reusable material has to be identified, translated, adapted, debugged, merged with other material and maintained, all of which can be very difficult and expensive processes. Therefore, for reuse to be effective, it has to be supported by a set of adequate knowledge base construction, editing and maintenance tools. This report describes an HPKB effort that built a variety of tools and infrastructure aimed at supporting the ontology and knowledge base construction process. All these tools are centered around the PowerLoom Knowledge Representation and Reasoning (KR&R) system (http://www.isi.edu/isd/ LOOM/PowerLoom), which is a highly expressive, logic-based KR&R system with multiple built-in deductive reasoning capabilities including a query processor, a description classifier, and a context mechanism. The developed tools cover various areas of the knowledge base and ontology construction process and are outlined in the report.