International Journal of Advanced Research in Computer Science and Software Engineering (IJARCSSE)
Many data mining techniques have been proposed with regard to mining valuable patterns with text docs. However, how the users' can effectively utilize and update discovered patterns is still an start research problem, especially from the domain regarding text mining. Since nearly all existing text message mining methods adopted term-based methods, they all experience the problems of polysemy and also synonymy. Over the years, people include often used the hypothesis that style (or phrase)-based methods should perform better than the term-based people, but a lot of experiments don't support this hypothesis.