针对互联网环境下新词出现和更新频率高的特点,将机械分词与基于规则分词相结合,提出一种动态更新词库的中文分词架构.本架构给出了新的词典设计结构及歧义处理规则,并将统计学中的互信息概念运用到新词判定环节.实验表明本文提出的中文分词架构具有较高的准确率和良好的适应性.
Aiming at new word appearing and updating frequently in the Internet, through combining mechanical segmentation and rule-based segmentation, a framework for Chinese word segmentation with dynamic updating dictionary is provided. We improve a new dictionary design structure and ambiguity processing rules. Also the concept of mutual information in Statistics is used in determining the new word. The experiment shows that the framework for Chinese word segmentation proposed in this paper has high precision and is adaptive in different domains.