Inductive Logic Programming (ILP) is the study of learning methods for data and rules that are represented in first-order predicate logic [Muggleton]. ILP methods mostly use logic programming as a uniform representation language for examples, background knowledge and hypotheses. Background knowledge holds the information about the language used to describe the examples and concepts, such as possible values of variables, hierarchies, predicates, and rules. ILP induces hypotheses from examples represented as first-order predicates and synthesize new knowledge from the examples. There are two standard approaches in ILP, one is bottom-up and second is top-down. Bottom-up programs implemented in systems such as ALEPH (A Learning Engine Processing Hypothesis) start with a very specific clause (also called a bottom clause) generated from a seed positive example and generalize it as far as possible without covering negative examples. The purpose of ILP is to discover definition of target predicates together with background knowledge such that it entails positive examples and not negative examples. The aim of this research is to implement a bottom-up learning mechanism incorporating a bottom clause for implementing Inductive Logic Programming methods using standard DBMS software to represent data and a Java interface to implement the ILP methods.
University of Minnesota M.S. thesis. 2017. Major: Computer Science. Advisor: Dr. Richard Maclin. 1 computer file (PDF); 70 pages.
Implementation of Breadth-First Search Method Based on a Randomly Chosen Bottom Clause for Inductive Logic Programming Method.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.