CFV-NB: Nave-Bayes Documents Classification Model Based on Concept Feature Vectors
Liu Jun · Jisuanji gongcheng · 2006
This paper proposes a novel Nave-Bayes document classification method based on the set of concept feature vectors. It produces someinitial classes from the set of unlabeled Web documents by SOM clustering and distributes a label for each, and builds the corresponding conceptfeature vector for each initial class using the maximum entropy method. It builds the last CFV-NB document classifier based on the space of conceptfeature vectors.