交通管理领域的中文分词算法及应用研究  

The Algorithm and Research of Chinese Word Segment in Intelligence Traffic Management Field

作　　者：熊桂喜[1] 姚丽[1] 

机构地区：[1]北京航空航天大学计算机学院,北京100191

出　　处：《微计算机信息》2009年第30期135-136,215,共3页Control & Automation

摘　　要：在分析智能交通管理(ITMS)领域特点的基础上,提出了适于ITMS的分词算法。使用特征词处理规则、专有词典和专有地名库切分出特征词、专有词和地名,应用N-最短路径法和基于一阶马尔可夫模型最小交叉熵方法做歧义处理,基于角色标注的方法识别人名,从N个最有潜力的候选结果中选优得到切分结果。在ITMS领域的测试集下分词系统的准确率和召回率分别达到96.3%和95.0%。This paper gives a segmentation solution based on the characteristics of ITMS. It separates signature word,specific words and Chinese place names in the first step. An n-shortest-path strategy and a method called minimum cross entropy based on 2- gram are employed to deal with segmentation ambiguity, and recognizes Chinese person names based on role tagging and get the final result from the top N segmentation candidates. The experiment shows that the precision and recall rate reach 96.3% and 95.0% respectively for ITMS Test Set.

关 键 词：智能交通管理 中文分词 人名识别 地名识别 专有词 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]