基于语义解析的领域问答系统的设计与实现  

DESIGN AND IMPLEMENTATION OF DOMAIN QUESTION-ANSWERING SYSTEM BASED ON SEMANTIC ANALYSIS

作　　者：刘园园 李劲华 赵俊莉 Liu Yuanyuan;Li Jinhua;Zhao Junli(School of Data Science and Software Engineering,Qingdao University,Qingdao 266071,Shandong,China)

机构地区：[1]青岛大学数据科学与软件工程学院,山东青岛266071

出　　处：《计算机应用与软件》2021年第11期42-48,97,共8页Computer Applications and Software

基　　金：国家自然科学基金项目(61702293);山东省重点研发计划项目(2019JZZY020101);中国博士后科学(2017M622137)。

摘　　要：针对高考咨询问题重复性高、搜索引擎回答较为冗余等现状,从各网站的半结构化以及非结构数据中抽取知识三元组,构建高校信息知识图谱;对涉及多种约束的复杂问句的问答方法进行研究,提出一个基于语义解析的问答模型,通过对问句的解析将问句自动转换为Neo4j查询语句。实验结果表明,该系统能较为准确地回答大部分领域问题,为其他领域类问答系统的实现提供了借鉴。Aiming at the current situation of high repetition of college entrance examination consultation questions and redundant search engine answers, knowledge triples from semi-structured and unstructured data of each website were extracted to construct a knowledge graph of university information. A question-answering model based on semantic analysis was proposed for complex questions involving multiple constraints. Questions were automatically converted into Neo4 j query sentences by analyzing the questions. The experimental results show that this system can answer most domain questions more accurately, and provides a reference for the implementation of other domain-based question answering systems.

关 键 词：知识图谱 问答系统 语义解析 高考咨询 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]