Marcella.Zaragoza@ISL1.RI.CMU.EDU.UUCP (01/30/87)
AI SEMINAR TOPIC: Knowledge-Based Reasoning at the Right Level of Abstraction: A Generic Task Toolkit SPEAKER: B. Chandrasekaran Laboratory for Artificial Intelligence Research Department of Computer and Information Science The Ohio State University Columbus, Ohio 43210 PLACE: Wean Hall 5409 DATE: Tuesday, February 3, 1987 TIME: 3:30 pm ABSTRACT: The first part of the talk is a critique of the level of abstraction of much of the current discussion on knowledge-based systems. It will be argued that the discussion at the level of rules-logic-frames-networks is the "civil engineering" level, and there is a need for a level of abstraction that corresponds to what the discipline of architecture does for construction of buildings. The constructs in architecture, viewed as a language of habitable spaces, can be @i(implemented ) using the constructs of civil engineering, but are not reducible to them. Similarly, the level of abstraction that we advocate is the language of generic tasks, types of knowledge and control regimes. In the second part of the talk, I will outline the elements of a framework at this level of abstraction for expert system design that we have been developing in our research group over the last several years. Complex knowledge-based reasoning tasks can often be decomposed into a number of @i(generic tasks each with associated types of knowledge and family of control regimes). At different stages in reasoning, the system will typically engage in one of the tasks, depending upon the knowledge available and the state of problem solving. The advantages of this point of view are manifold: (i) Since typically the generic tasks are at a much higher level of abstraction than those associated with first generation expert system languages, knowledge can be represented directly at the level appropriate to the information processing task. (ii) Since each of the generic tasks has an appropriate control regime, problem solving behavior may be more perspicuously encoded. (iii) Because of a richer generic vocabulary in terms of which knowledge and control are represented, explanation of problem solving behavior is also more perspicuous. We briefly describe six generic tasks that we have found very useful in our work on knowledge-based reasoning: classification, state abstraction, knowledge-directed retrieval, object synthesis by plan selection and refinement, hypothesis matching, and assembly of compound hypotheses for abduction. Finally, we will describe how the above approach leads naturally to a new technology: a toolbox which helps one to build expert systems by using higher level building blocks. We will review the toolbox, and outline what sorts of systems can be build using the toolbox, and what advantages accrue from this approach.