Task-based explanations

Abstract Explaining a knowledge system's conclusions requires, among other things, showing how it has satisfied the logical requirements of its problem-solving task. In one report it was shown how task-based explanation worked in generic tasks (GT); other GT work showed how more complex tasks could be solved using multiple GT modules. Because each GT in a composite GT knowledge system serves a role defined by the higher-level goals it serves (i.e., a GT module is a method that achieves a particular goal of the higher-level task) it is possible to explain the high-level task in terms of the low-level GT explanations. We have introduced the idea of a logical structure for diagnosis, a task solved by composition of several GTs, and its use in explanation. In this article we describe how these ideas work to generate explanations in a composite GT system.

[1]  William B. Thompson,et al.  Reconstructive Expert System Explanation , 1992, Artif. Intell..

[2]  Gary S. Kahn,et al.  The Mud System , 1986, IEEE Expert.

[3]  David C. Brown,et al.  Design Problem Solving: Knowledge Structures and Control Strategies , 1989 .

[4]  B. Chandrasekaran,et al.  Design Problem Solving: A Task Analysis , 1990, AI Mag..

[5]  M. C. Tanner.  Explaining knowledge systems: justifying diagnostic conclusions , 1989 .

[6]  J W Smith,et al.  RedSoar--a system for red blood cell antibody identification. , 1991, Proceedings. Symposium on Computer Applications in Medical Care.

[7]  Luc Steels,et al.  Components of Expertise , 1990, AI Mag..

[8]  Dean Allemang,et al.  The Computational Complexity of Abduction , 1991, Artif. Intell..

[9]  Luc Steels,et al.  Second-Generation Expert Systems , 1985, IEEE Expert.

[10]  J. Reggia,et al.  Abductive Inference Models for Diagnostic Problem-Solving , 1990, Symbolic Computation.

[11]  W. F. Punch,et al.  An investigation of the roles of problem-solving methods in diagnosis , 1993 .

[12]  Ashok K. Goel,et al.  Structured matching: a task-specific technique for making decisions , 1991 .

[13]  Allen Newell,et al.  SOAR: An Architecture for General Intelligence , 1987, Artif. Intell..

[14]  Todd R. Johnson,et al.  Task-structure analysis for knowledge modeling , 1992, CACM.

[15]  B. Chandrasekaran,et al.  Generic Tasks in Knowledge-Based Reasoning: High-Level Building Blocks for Expert System Design , 1986, IEEE Expert.

[16]  B. Chandrasekaran,et al.  A Mechanism for Forming Composite Explanatory Hypotheses , 1987, IEEE Transactions on Systems, Man, and Cybernetics.

[17]  Michael C. Tanner.  Explanation in Generic-Task Systems , 1992, AIFIPP.

[18]  Raymond Reiter,et al.  A Theory of Diagnosis from First Principles , 1986, Artif. Intell..

[19]  Sanjay Mittal,et al.  CRSL: A Language for Classificatory Problem Solving and Uncertainty Handling , 1986, AI Mag..

[20]  William J. Clancey,et al.  NEOMYCIN: Reconfiguring a Rule-Based Expert System for Application to Teaching , 1981, IJCAI.

[21]  Sandra Marcus,et al.  Automating Knowledge Acquisition for Expert Systems , 1988 .

[22]  William J. Clancey,et al.  Heuristic Classification , 1986, Artif. Intell..