Working Notes of the 1990 Spring Symposium on Automated Abduction
Abstract
The philosopher Charles Sanders Pierce used the term abduction for a form of inference considered to be as important as deduction and induction. Abduction is concerned with explanatory reasoning and is closely related to the relatively modern notions of backward chaining and inference to the best explanation. Since explanations are important in many different aspects of intelligence, cognitive scientists have become interested in computer programs that construct and evaluate explanations. In artificial intelligence, a number of key tasks have come to be viewed in terms of abduction. In expert systems, the best known abduction problem is diagnosis. In natural language comprehension, plan recognition is viewed as an abduction problem involving the inference of goals from observed behavior. In qualitative physics, postdiction is an abduction problem involving explaining states of the physical world in terms of processes and casual laws. In machine learning, explanation-based learning strategies improve performance using processes that construct explanations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 27, 1990
- Accession Number
- ADA228005
Entities
People
- Paul O'rorke
Organizations
- University of California, Irvine