With Distance From The Automatic Word Clause

With the progress of natural language applications, it is more and more important to improve the validity of Chinese word-split and sentence-split, which is the basic of natural language process. Researching the automatic judge of Chinese word-split and sentence-split becomes an urgent task.At the present time, there are parts of automatic judge which are put forward by academic. But there are some disadvantages in these methods. For example most of these methods only allow a part of words or assembles and the percentage of correctness is low. In a word, researching the automatic judge of Chinese word-split and sentence-split is just on the drawing boards.So on the base of research and analyses of the technology and method of the word-split,sentence-split and automatic judge, referring to the thinking pattern when men meet similar problem, the semantic repositories are presented by the paper as a breach. The paper tries to find a assemble measurement with the help of relation cobweb of Chinese words.A new idea, assemble distance, is presented by the paper with the support of HowNet and the framework of semantic distance, based on the research and analyses of the primary semantic repositories.The paper will discuss how to define assemble distance and how to calculate it with some instances. It is clearly that assemble distance can effectively measure the validity of assembles. It can generate a goodish result combined with other method of Chinese word-split and sentence-split, which is proved by experiments in the paper. The experiments also bring an exceeding result of measurement to simple various meanings sentences and solecism.In the last part of the paper, a work plan for future to improve the percentage of assemble distance is marked out.