基于多种知识源的汉语自动分词

A Chinese Word Segmenting Scheme Based on Multiple Knowledge Sources

提出一种汉语分词方法。与其它的如利用单一统计特性的统计方法或者纯规则方法不同，该方法利用字、词、上下文、语法及语义等多种知识源对汉字串中每一隔点的切分可能性进行考察，并在无法彻底消歧的情况下通过模糊综合得出最可能的切分结果。用户可以根据需要修改系统以适应不同文本的特征，并能接收前后词法、语法、语义分析阶段的反馈。因此，该方法具有准确率高、灵活、健壮、回溯迅速的特点。

A new method for Chinese word segmenting based on multiple knowledge sources is proposed The segmenting possibility at each segmentable point of the Chinese character string is figured out by taking advantage of multiple knowledge sources, and the most possible segmenting path is found out by fuzzy synthesis