Semantic analysis of inductive reasoning (Q1081317)

A semantic model of inductive reasoning is presented. The starting point is a general schema: ''given background knowledge K and an observational statement S, a sentence H is an inductive hypothesis for S with respect to K iff H is a generalization of S, and H is consistent with K''. For semantic modelling, a set-theoretic approach is adopted: the universe of discourse is a fixed set of objects with attributes, the background knowledge is a set of binary relations between objects and the concept is modelled by ''an approximate set'' represented as a pair of object sets which specify the limits of concept definition tolerance. Conclusively, the inductive inference of concepts from examples and counterexamples is defined. With some improvements depending on concrete application, such a model could be useful in machine learning area, as well as in knowledge based computer systems.