Abstract: "Some research [3, 1, 2] has revealed logical aspects of analogy based generalization. Given similarity S and projected property P, inductive generalization gives a rule [for all] x. S(x) [is a superset of] P(x) from a fact [there exists some] x. S(x) [and] P(x). In this paper, we sho...

Ausführliche Beschreibung

Bibliographische Detailangaben

Hauptverfasser:	Iwayama, Noboru (VerfasserIn), Satoh, Ken 1959- (VerfasserIn), Arima, Jun (VerfasserIn)
Format:	Buch
Sprache:	English
Veröffentlicht:	Tokyo, Japan 1992
Schriftenreihe:	Shin-Sedai-Konpyūta-Gijutsu-Kaihatsu-Kikō <Tōkyō>: ICOT technical report 735
Schlagworte:	Logic programming
Zusammenfassung:	Abstract: "Some research [3, 1, 2] has revealed logical aspects of analogy based generalization. Given similarity S and projected property P, inductive generalization gives a rule [for all] x. S(x) [is a superset of] P(x) from a fact [there exists some] x. S(x) [and] P(x). In this paper, we show that a principle for analogy based on generalization is described and implemented in logic programs very clearly and simply. By the description, we give a declarative semantics of analogy based generalization. We also consider the implementation as being an analogical system by itself to support research on the analogy. We need integrity constraints to describe the intrinsic non-deductiveness of the analogy We consider the conclusion of analogical inference as one of minimal models which are not generally unique for a program. Therefore, stable model semantics is suitable for our formulation of the analogy. So, from the aspect of logic programming, this work gives us an important application of integrity constraints and a fine instance of utilization of stable model semantics. We show the correspondence between an analogical model for a Horn logic program with integrity constraints and a stable model for an extended program of the original program. We also show the computation of the analogy by non-deterministic procedure of stable models.
Beschreibung:	14 S.