一种面向网络答疑的汉语切分歧义消除算法  

A Novel Algorithm of Eliminating the Chinese Word Segmentation Ambiguities for Web Answer

作　　者：张素娟[1] 郑庆华[1] 胡云华[1] 孙霞[1] 

机构地区：[1]西安交通大学计算机系,西安710049

出　　处：《计算机工程与应用》2004年第25期55-58,共4页Computer Engineering and Applications

基　　金：国家自然科学基金项目(编号:60373105);"十五"重大科技攻关项目(编号:2001BA101A01);教育部优秀青年教师基金项目资助

摘　　要：针对网络答疑的特点,该文提出了一种汉语歧义消除算法,采用回溯机制及歧义消除评优算法相结合的方法消除汉语切分歧义。首先利用回溯机制发现句子的切分歧义字段,产生含有多种可能切分结果的候选集;然后针对网络答疑特点提出了评优算法,利用该算法计算候选结果的评价值,对其进行排序,选取分词最佳结果,从而消除歧义,提高分词的准确率。该算法已经在基于Web的自然语言答疑系统WebAnswerSystem中实现并得到了实际应用。实验结果表明,算法具有较高的准确率和召回率,对自然语言网络答疑中进行切分歧义消除是行之有效的。Motivated by the requirement of Web Answer for distance learning,a novel algorithm of eliminating the ambiguity of words segmentation for Web Answer is proposed in this paper.Firstly,we take a back tracking mechanism to find possible results of the word segmentation to make the candidates,then we propose a optimal judgment algorithm according to the characteristic of the Answering texts,finally This algorithm is used to choose the most suitable one from all candidates and eliminate the ambiguity of words segmentation.The proposed algorithm has already been applied in our Web answering system,the experimental result shows that it has high accuracy and efficiency of word segmentation and elimination of the ambiguity of words segmentation.

关 键 词：网络答疑 切分歧义 回溯机制 评优算法 

分 类 号：TP391.12[自动化与计算机技术—计算机应用技术]