任意输入字符串，把相同的字符清楚掉，剩下的字符按原先的排序组合在一起-Any input string, the same character clear out the rest of the characters are combined together in the original sorting

对于大文本进行挖掘聚类，该方法不考虑文字词语出现的频率信息，考虑上下文语境，将所有的字根据预定义的特征进行词位特征学习，获得一个训练模型。然后对待分字符串的每一个字进行词位标注，最后根据词位定义获得最终的分词结果。(Digging for large text clustering, the method does not consider the text word frequency of information, considering the context, all the words