A Combined Weight method in Automatic Classiflcation of Chinese Text

In this paper, we set a shielded level in a concept tree to use both the concept attributes from a semantic dictionary and the Chinese words to make the feature set. After comparing the weight theories and classification precise, of the eight methods, we give a new selection method, the CHI-MCOR weight method, which is derived from two normal methods which present well in our experiments. Our former experiment result shows that if we can set a proper shielded level, we can not only reduce the feature dimension but also improve the classification precise. The later result shows that the combined weight method makes a good balance between the fuzzy words which have a high occurrence and the dividing words which have a middle or low occurrence, and the classification precise is higher than any one of the weight methods