Combining fuzzy clustering with Naive Bayes augmented learning in text classification

For obtaining labeled training samples in text data mining, transcendental knowledge of samples and non-supervisory of clustering were combined. Fuzzy Partition Clustering Method (FPCM) was presented and used to obtain a few labeled texts and some external clusters automatically by measuring the similarity degree of clustering correlation texts. So classification bases were found for supervised learning. Naive Bayes augment learning manner was further combined to design and learn classifiers, and the way of estimating the loss of classing error was used to balance the selection of those example candidates. The combination of those two methods has advanced the precision of text classification and makes classification learning of non-labeled training example with more potential applications.