摘要
自动应答系统中对用户所提问句的理解是系统实现的关键,同样也是一个难点,通过在受限领域内建立问句语料库来协助理解用户问句是一种非常有效的实现方法。文章分析了建行领域业务咨询系统的问句收集、分词和词性标注、语义标注、问句语料统计等问句语料库的建设过程,并详细介绍了采用词向量空间法和语义向量空间法从问句语料库中寻找和目标问句相似问句的计算方法及提取答案的实现过程。
Understanding user's question sentences is the key problem in question answer system.That's also the diffi-cult part in the whole system.It is an effective method to understand user's intention by constructing question sentence corpus.This paper analyzes the process of constructing question sentence corpus in building consulting system for china construction bank,including question sentence collection,word segmentation and Pos tagging,Semantic tagging,and ques-tion sentence corpus statistics,and introduces an algorithm to find similar sentence with target question sentence from question sentence corpus and extract the answer by using Word Vector Space Method and Semantic Word Vector Space Method in details.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第36期28-30,86,共4页
Computer Engineering and Applications
基金
云南省基金项目资助(编号:2002IT03)
关键词
自然语言处理
问句语料库
自动应答系统
问句语义标注
Restricted Domain Question Answer System(QAS),Question Sentence Corpus,Semantic tagging,Vector Space Model,Semantic similarity,Question sentence similarity