15th European Conference on Artificial Intelligence
|
July 21-26 2002 Lyon France |
[full paper] |
Patrick Doherty, Witold Lukaszewicz, Andrzej Szalas
We introduce a Computer-Aided Knowledge Engineering (CAKE) technique for incrementally constructing, managing and querying knowledge structures. These knowledge structures can be used to represent incomplete or approximate world models used for sofbots or robots to reason with in addition to being used for applications on the semantic web. Conceptually, the knowledge structures represent a generalization of deductive databases where relations and properties are approximate in nature and the query mechanisms can be contextualized to locally close different parts of the database under an open-world assumption. Pragmatically, the knowledge structures are represented using a diagrammatic technique with a formal semantics. A query to a knowledge structure can easily be compiled into a query to a rough relational database. The underlying semantics and computation mechanism insures that any reasoning expressed by the CAKE diagrams is computable in deterministic polynomial time. The diagrams themselves can be viewed as confederations of granular agents, each responsible for managing a relation or property, or default rules and their adjudication under conflict. This view contributes to the incremental construction, modularity and compositionality of the technique which is demonstrated in the paper.
Keywords: Knowledge Representation, Common-Sense Reasoning, Nonmonotonic Reasoning, Deduction, Knowledge-Based Systems
Citation: Patrick Doherty, Witold Lukaszewicz, Andrzej Szalas: CAKE: A Computer-Aided Knowledge Engineering Technique. In F. van Harmelen (ed.): ECAI2002, Proceedings of the 15th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2002, pp.220-224.