A Genetic Algorithm for Discovering Interesting Fuzzy Prediction Rules: applications to science and technology data

Data mining consists of extracting knowledge from data. This paper addresses the discovery of knowledge in the form of prediction IF-THEN rules, which are a popular form of knowledge representation in data mining. In this context, we propose a new Genetic Algorithm (GA) designed specifically for discovering interesting fuzzy prediction rules. The GA searches for prediction rules that are interesting in the sense of being surprising for the user. More precisely, a prediction rule is considered interesting (or surprising) to the extent that it represents knowledge that not only was previously unknown by the user but also contradicts the original believes of the user. In addition, the use of fuzzy logic helps to improve the comprehensibility of the rules discovered by the GA, due to the use of linguistic terms that are natural for the user. The proposed GA is applied to a real-world science & technology data set, containing data about the scientific production of researchers. Experiments were performed to evaluate both the predictive accuracy and the degree of interestingness (or surprisingness) of the rules discovered by the GA, and the results were found to be satisfactory. 1