Abstract
In this paper, we propose a rule extraction system for ordinal scale data using logic minimization technique. Logic minimization is a heuristic optimization method for reducing the number of logic circuits in VLSI design. Logic minimization technique corresponds to rule extraction from decision tables that enable to explain many sample data by small rules. However, logic minimization does not treat ordinal relations among attribute values and is not suitable for extracting rules from ordinal scale data. We then expand logic minimization technique to treat order relations among attribute values and verify the usefulness of the proposed rule extraction for ordinal scale data by experiments.
