Research on Feature Extraction Based on Genetic Algorithm in Text Categorization

Using the superiority of the genetic algorithms solving nonlinear problems which is applied to feature extraction on Text Categorization is proposed in this paper. A synonym problem in text is fully considered by this method during the feature extraction processing, and is processed using the concept of fuzzy set. The membership of synonymous as the fitness function of Genetic Algorithm is carried out by evolutionary computation. Comparison test results show that this method not only can reduce the dimension of feature, but also can improve accuracy ration and recall ratio of classification, the overall performance of the classification system is improved finally, the system is achieved a higher level of automation and the strong portability.