Buchtitel: Proceedings of the 10th European Workshop on Knowledge Acquisition, Modeling, and Management (EKAW'97)
Reihe: Lecture Notes in Artificial Intelligence (LNAI)
We present a structured approach for developing problem-solving methods for knowledge-based systems. We view this process as a stepwise combination process of methods and adapters. We start from very generic search strategies with very general data structures and add adapters that refine the states of the search process, that refine state transitions, and that add assumptions necessary to link the competence of a method with given problem definitions and domain knowledge. The paper provides three novel contributions to knowledge engineering. First, we provide a structured approach for the development and adaptation of problem-solving method. Second, we show how the usability-reusability trade-off of task-specific versus task-independent problem-solving methods can easily be overcome by the virtual existence of specific methods. Third, we provide the concept of an integrated library combining reusable problem definitions, problem-solving methods, and adapters.