银行领域汉语自动问答系统BAQS的研究与实现

A Study on a Bank-Domain Automatic Chinese Question-Answering System BAQS

  • 摘要: 介绍BAQS的研究背景和系统框架.探讨基于问点块和语义块识别以及句模匹配分析问句的新方法,并用向量表示整个问句语义.借鉴本体和知网思想,构建银行领域本体库和银行知网.采用预先对金融领域实用文本进行标注,依据问句向量从标注树中提取答案.并针对某银行实现汉语自动问答系统.实验表明该方法可行,对自动问答系统的设计具有借鉴意义和深入研究的价值.

     

    Abstract: The background and system frame of the system BAQS are presented. A new approach is explored in which questions are analyzed based on the detection of question focus chunk, semantic chunk and question template. Then vectors are used for the whole semantic representation. With the elicitation of ontology and Hownet, a bank ontology database and bank Hownet are built up. The finance domain texts are annotated in the head, and then the answer extracted from the tag trees by the question vector. For a specific bank a question-answering system is implemented and experiments do prove that it is feasible to use the method to develop a domain question-answering system, that is valuable for further study in more depth.

     

/

返回文章
返回
Baidu
map