Extending inductive generalization with abduction
Takashi Kanai, Susumu Kunifuji · Institutional Repositories DataBase (IRDB) · 1999
We proposes an integrated framework of inductive generalization and abductive reasoning. In this framework, inductive hypotheses can be generated even if the background knowledge is not sufficient to generate hypotheses in an usual manner. The main issue of inductive generalization is to construct definitions of given examples when examples and relevant background knowledge are given. While most inductive generalization systems presuppose that the background knowledge is enough to induce definitions of given examples, the assumption is not usually met in real-world situation. In order to overcome this difficulty, we propose a framework of inductive generalization extended with abductive reasoning. This approach uses abduction when inductive generalization needs more items of background knowledge. Our approach is an integration of Inductive Logic programming and Abductive Logic Programming.