Abstract: | If we can express a domain theory in a logical formalism and formulate a task in that domain in mathematical terms, then this greatly facilitates the task of building sound knowledge based systems. However, it is not immediately obvious how the design aspects of such tasks, where these are an important feature of problem-solving, can be incorporated in this approach. We have used a methodology based on the development of proof plans. The ability to conduct our research according to the Popperian framework of hypothesis, validation, testing, and modification in response to empirical evidence - the hypothetico-deductive approach - seems essential to any rigorous scientific endeavour. Proof planning, we believe, is a method which readily exploits this inherently incrementalist approach, and could prove a powerful tool in designing AI systems.
|