Micro-blog category based on feature-words category dispersion

The micro-blog information classification is an important pretreatment in micro-blog data processing work. Due to the unique properties of the micro-blog text, there are some limitations when use traditional classification to deal with it. Consider to a single microblog text brief which contains less effective feature-words, and the content compare spoken of the features, this paper proposed to use similar words and collocations to extend the text feature-words, reducing the possibility of feature loss. For the feature of information selection and weight calculation, proposed one kind text classification methods which based on the feature-words category dispersion and dispersion degree. The experiments show that the propose classification method achieves good effects in the classification of micro-blog text, and has better applicability in microblog text classification scene.