Interactive transfer of expertise: Acquisition of new inference rules

Classics

Summary of Ph.D. dissertation, Computer Science Dept., Stanford University (1979)."TEIRESIAS is a program designed to provide assistance on the task of building knowledge-based systems. It facilitates the interactive transfer of knowledge from a human expert to the system, in a high level dialog conducted in a restricted subset of natural language. This paper explores an example of TEIRESIAS in operation and demonstrates how it guides the acquisition of new inference rules. The concept of meta-level knowledge is described and illustrations given of its utility in knowledge acquisition and its contribution to the more general issues of creating an intelligent program."Also in:Readings in Artificial Intelligence, ed. Webber, Bonnie Lynn and Nils J. Nilsson, Palo Alto, CA: Tioga Publishing Co., 1981.Orig. in IJCAI-77, vol.1, pp. 321 ff. Preprint in Stanford HPP Report #HPP-77-9.See also: Artificial Intelligence, 12[#2]:409-427. Readings in Artificial Intelligence, ed. Webber, Bonnie Lynn and Nils J. Nilsson, Palo Alto, CA: Tioga Publishing Co., 1981


On Competence and Meta-Knowledge Gerhard Wickler Louise Pryor

AAAI Conferences

Department of Artificial Intelligence University of Edinburgh 80 South Bridge Edinburgh EH1 1HN Scotland {gwllouisep}@aisb.ed.ac.uk Abstract In this paper we define and attack the problem of competence assessment for intelligent agents. The basic idea is that we use metaknowledge to infer competence. The main contribution of this paper is a single rule that allows efficient competence assessment for any system with explicit strategic knowledge. The reason for this is that strategic knowledge already contains the right information. Cognitive evidence supports our theory. Competence and Intelligent Agents The Problem of Competence Assessment The problem we attempt to address in this paper is best illustrated by looking at an example. Consider the problem-solving activity of human problem solvers given the following simple physics problem1: A block of mass m starts from rest down a plane of length l inclined at an angle O with the horizontal. If the coefficient of friction between block and plane is #, what is the block's speed as it reaches the bottom of the plane? Given that the human problem solvers have some knowledge of physics in the form of equations that are appropriate to the problem, they will most likely answer the question whether they can solve this problem with "yes", i.e. they will state that they are competent to solve this particular problem instance.


William J. Clancey

AI Magazine

Origins The idea of developing a tutoring program from the MYCIN knowledge base was first described by Ted Shortliffe (1974). In fact, it was the mixed-initiative dialogue of the SCHOLAR teaching program (Carbonell, 1970) that inspired Shortliffe to produce the consultation dialogue of MYCIN. He conceived of it as a question-answer program in SCHOLAR's style, using a semantic network of disease knowledge. Shortly after I joined the MYCIN project in early 1975, Bruce Buchanan and I decided that developing a tutoring program would be my thesis project. The GUIDON program was operational in early 1979.


Planning and Meta-Planning

Classics

Summary of PhD thesis, Computer Science Department, Stanford University, January, 1980,Stanford Rep. Nos. HPP-80-2, STAN-CS-80-784


The GAMES-II Methodology for medical KBS development

AAAI Conferences

While the first one is aimed at selecting a set of hypotheses representing possible solutions of the problem at hand, the second one is aimed at testing each of them. Because of the paradigm on which it is rooted, the model has been dubbed select and test model (STModel). It can represent the three generic medical tasks: diagnosis, therapy planning and patient monitoring, by defining the knowledge roles played by the domain entities in each one of them. The GAMES-II methodology is based on three principles that have emerge during the last decade of AI research: (i) knowledge level modeling, (ii) reusability of both task and domain knowledge, (iii) integration of multiple reasoning techniques. The knowledge modeling (Newell, 1982) principle states that knowledge should be modeled on a higher level than that of exploited knowledge representation formalisms, to avoid premature design decisions, and to facilitate communication with domain experts. The second principle implies that the complexity of KBS development must, just as with any other engineering activity, be tackled by the construction of libraries of reusable components (Puerta et al., 1992). The third principle is based on the consideration that the "weak methods" are too weak, and therefore that KBS should use multiple specialised reasoning techniques for the different steps in the problem solving process.