用B-树作为查找存储结构，对中文单词进行哈希，用最长逆向匹配算法进行中文句段分词，实现中文词典以及分词。中文词典规模为十万八千多词汇。分词的正确率在90 以上。-Use the B-tree as storage structure , and hash the Chinese word while storing or searching. Use the longest reverse matching algorithm to split Chinese　sentence to words. Chinese dictionaries scale of more than 108,000 words. Word correct rate of over 90 .