Assembling the best explanation

Authors:
John R. Josephson;B. Chandrasekaran;Jack W. Smith, Jr.
Affiliations:
Department of Computer and Information Science and Department of Pathology, The Ohio State University, Columbus, Ohio;Department of Computer and Information Science and Department of Pathology, The Ohio State University, Columbus, Ohio;Department of Computer and Information Science and Department of Pathology, The Ohio State University, Columbus, Ohio
Venue:
PKWBS-W'84 Proceedings of the 1984 IEEE conference on Principles of knowledge-based systems
Year:
1984

Citing 3
Cited 0

The formation of composite hypotheses in diagnostic problem solving: an exercise in synthetic reasoning

IJCAI'77 Proceedings of the 5th international joint conference on Artificial intelligence - Volume 2
DIALOG: a model of diagnostic logic for internal medicine

IJCAI'75 Proceedings of the 4th international joint conference on Artificial intelligence - Volume 1
On the mechanization of abductive logic

IJCAI'73 Proceedings of the 3rd international joint conference on Artificial intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

Going from data describing a situation to an explanatory hypothesis that best accounts for the data is a commonly occurring knowledge-based reasoning problem. Its presence can be detected in such diverse tasks as diagnosis, perception, and theory formation. This form of reasoning has been called "abductive inference"1, 2 and we adopt the term here. Sometimes in order to accomplish an abduction the need is to assemble hypothesis parts into a unified explanatory hypothesis. In this paper we describe a general mechanism for accomplishing the unification of sub-hypotheses with possibly overlapping domains of explanation. This mechanism makes use of plausibility information concerning the sub-hypotheses, along with information about what a sub-hypothesis can explain in the particular situation, to build up a best explanation. The novel capability arises of "abductive confirmation" of a sub-hypothesis based on its ability to explain some feature for which no other plausible explanation can be found. A version of this mechanism has been used successfully as the basis for a knowledge-based system, RED, which solves real-world problems of red-cell antibody identification3. These are problems which arise in the hospital blood bank, and are currently solved by specially trained human experts.