Study On Mobile Phone Automatic Classification Based On The Improved K-nearest Neighbor Algorithm

In recent years, smart mobile phone with the rapid development, gradually become a new media forms followed by television, radio, newspapers, Internet. Sending and receiving Email through by smart mobile phone is much more frequent than ever, combined with the spam Email diffusion which we have not curt and become more and more dangerous. Moreover, the Email providers constantly enhance the capacity of the mailbox by geometrical speed. So when facing up with a flood of mail and spam Email’s frequent incursions, it is important and crucial for Email automatic classification.The phone Email automatic classification based on Email content belongs to an important application of data mining in Chinese text classification, so this article firstly introduces the Email parsing, the text classification techniques and some background of automatic text classification, and then talking about the process of automatic text classification, finally studying the Chinese email text automatic classification system involved in various aspects of the theory and technology. We not only study the Key technology for text classification, including the Chinese word segmentation, feature extraction, word frequency statistics, vector space model and classification algorithms, but also theoretical explanations and algorithm description.This article focuses on the K-nearest neighbor algorithm, due to the blindness of selecting K threshold in K-nearest neighbor algorithm, we come up with improved K-nearest neighbor algorithm. This algorithm is based on the singular value decomposition of a matrix technology, which can be able to quickly get through the rough points and the breakdown of two stages of training samples and the estimated value of K, then by getting the training sample to train K-nearest neighbor classifier, and also implementing an automatic Email classifier in Java language. Finally, we judge and verify this classifier from recall and precision of those two aspects.