Classifying web pages and documents based on expected cross entropy and weighted vote schema

Traditional information retrieved method use keywords occurring in determine the class of the documents and web pages, but usually retrieves unrelated web page and documents. We propose a web pages and documents scanning and classification method base on support vector machine and expected cross entropy and using a weighted vote schema. Experimental results indicate our method is more effective than traditional methods. Classification accuracy in proposed method is better than other methods and even with a small labeled training set, our method could achieve higher accuracy.