Obtaining complex linguistic rules from decision information system based genetic algorithms

As a rule, in real decision information system (DIS), attributes\´ values of objects are continuous numerical. And it is difficult to "read" good information from the data. Numerical values are too fine to directly obtain decision rules which are expressed by natural language. In this paper, based on genetic algorithms (GA), extracting linguistic decision rules from DIS is discussed. In order to obtain complex linguistic decision rules, a fitness function for linguistic terms is proposed, the function expresses the degree of all the thresholds used in the sentences for each linguistic term, obtain complex linguistic decision rules.