Abduction Compared with Negation by Failure.

摘要: Horn clause logic programming can be extended to include abduction with integrity constraints. In the resulting extension of programming, negation by failure simulated making negative conditions abducible and imposing appropriate denials disjunctions as This gives an alternative semantics for failure, which generalises stable model failure. The abductive extends in three ways: (1) computation perfonned minimal models, (2) positive well made abducible, (3) other constraints also accommodated. * paper was written while first author at Imperial College. 235 Introduction tenn "abduction" introduced philosopher Charles Peirce [1931] refer a particular kind hypothetical reasoning. simplest case, it has fonn: From A fB infer B possible "explanation" A. Abduction been given prominence Charniak McDennot's [1985] "Introduction Artificial Intelligence", where applied expert systems story comprehension. Independently, several authors have developed deductive techniques drive generation hypotheses. Cox Pietrzykowski [1986] construct hypotheses from "dead ends" linear resolution proofs. Finger Genesereth generate "deductive solutions design problems" using "residue" left behind Poole, Goebel Aleliunas [1987] use All impose restriction that should consistent "knowledge base". is fonn non-monotonic reasoning, because are one state knowledge base may become inconSistent when new added. Poole [1988] argues preferable noh-monotonic logics default this view, defaults fonnulated within classical rather than conclusions derived withln some logic. similarity between reasoning pointed out [Kowalski, 1979]. we show how integrated concentrate on generalise Conditional Answers Compared program consists set Clauses, used backward to_reduce goals sub goals. initial goal solved there no subgollls left;