Integration of Learning into a Knowledge Modeling Framework

Enric Plaza & Josep Lluís Arcos

IIIA - Institut d'Investigació en Intel.ligència Artificial,
CSIC - Spanish Scientific Research Council,
Campus Universitat Autonòma de Barcelona,
08193 Bellaterra, Catalonia, Spain.
Voice: +34 3 5809570;
Fax: +34 3 5809661;
Email: plaza@iiia.csic.es & arcos@iiia.csic.es
WWW: http://www.iiia.csic.es


Abstract

In this paper we will report our current research on the NOOS language, an attempt to provide a uniform representation framework for inference and learning components supporting flexible and multiple combination of these components. Rather than a specific combination of learning methods, we are interested in an architecture adaptable to different domains where multiple learning strategies (combinations of learning methods) can be programmed. Our approach derives from the knowledge modelling frameworks developed for the design and construction of KBSs based on the task/method decomposition principle and the analysis of knowledge requirements for methods. Our thesis is that learning methods are methods with introspection capabilities that can be also analyzed in the same task/method decomposition. In order to infer new decisions from the results and behavior of other inference processes, those results and behavior have to be represented and stored in the memory for the learning method to be able to work with them.