Inductive logic programming: Theory and methods. (English) Zbl 0816.68043

The paper is an interesting and clear survey of the theory and the applications of inductive logic programming (ILP). This is a new discipline arising from the integration of inductive machine learning and logic programming. Similarly to the case of inductive learning, the aim of ILP is to develop techniques for constructing inductively hypotheses from examples and background knowledge. From logic programming ILP inherits the representation mechanism, by modeling knowledge in terms of first order clausal formulas. As argued by the authors, such an integration results in several benefits for the field of inductive inference in general. Moreover ILP contributes also to the theory of computational logic, by investigating induction rather than deduction as the basic mode of inference. A section on applications, ranging from drug design to satellite diagnosis and logic programs synthesis, concludes the paper and provides a convincing argument in favour of ILP.


68N17 Logic programming
68T05 Learning and adaptive systems in artificial intelligence
Full Text: DOI