Evolutionary data mining applied to TV databases : a first approach

Abstract

Recently, Evolutionary Algorithms (EAs) have been applied with very good results to various types of data mining problems. This paper presents the initial results of our work on data mining from TV program databases using EAs. We compare the performance of different operators and different operator parameters. INTRODUCTION Evolutionary Algorithms (EAs) have already proved their usefulness in scientific and real-world problems quite successfully. They are stochastic search techniques that explore combinatorial search spaces using simulated evolution. The primary objective of an EA is either to find something -- whether this is known or not -- or accomplish a goal, or, more generally, to solve a problem. EAs maintain a population of individuals that evolve according to a set of rules regarding selection, recombination and mutation. The population is evolved towards the optimum, concentrating search in those areas of higher fitness (exploitation). Recombination and mutation perturb the.