Text classification is a very important technique in the field of text mining

and it has been widely applied to the information management

search engine

and recommendation systems. Most existing classification methods cannot be used on the occasion of classifying a large number of samples. Therefore

a novel improved genetic algorithm was proposed for Web text mining system. The performance of the proposed system has been improved largely by our improved genetic algorithm. The proposed approach can be applied to classify a large number of samples. Experimental results show that both the accuracy and the speed of categorization are high.