Fuzzy Rule Miner: A Software Library Used in Project Based Teaching of Topics Related to Knowledge Discovery in Databases

Abstract

A group of fuzzy rules is a human-interpretable knowledge representation which is used in various fields of study such as control and knowledge discovery in databases, e.g. for inferring on output based on input variables. It makes use of the notions of fuzzy logic such as fuzzy sets, membership functions and membership degrees. Truth values represented by membership degrees may be any real number between 0 and 1. Effective definition of membership degrees through membership functions for particular applications requires a tool able to set membership functions and to generate inputs for various knowledge discovery algorithms. Similarly, various algorithms for making fuzzy rules and validating are required as well. In addition, modifications have to be possible as research in this scientific area is active. Since many fuzzy rule related algorithms and their modifications are not available for development of programs, software library Fuzzy Rule Miner is presented in this paper. This software library is written in Java and it can be used for helping students with fuzzy logic related calculations, fuzzy rule discovery and the use of fuzzy rules. It can also serve as a support library for complex software programs working with fuzzy rules.