基于百年党史知识图谱的智能问答系统设计与实现  

Design and Implementation of an Intelligent Q&A System Based on the Knowledge Graph of 100 Years of Party History

作　　者：杨思航 郑晓梅 李刚[1,2] Yang Sihang;Zheng Xiaomei;Li Gang(School of Artificial Intelligence and Information Technology,Nanjing University of Chinese Medicine,Nanjing Jiangsu 210023;Jiangsu Intelligent Chinese Medicine Health Service Engineering Research Center,Nanjing Jiangsu 210023;School of Cyber Science and Engineering,Nanjing University of Science&Technology,Wuxi Jiangsu 214443)

机构地区：[1]南京中医药大学人工智能与信息技术学院,江苏南京210023 [2]江苏省智慧中医药健康服务工程研究中心,江苏南京210023 [3]南京理工大学网络空间安全学院,江苏无锡214443

出　　处：《情报探索》2025年第2期42-50,共9页Information Research

基　　金：江苏省社会科学基金课题“基于语义关联的图博档融合知识图谱构建研究”(项目编号:21TQB001);江苏省教育科学“十四五”规划重点课题“数字赋能中医药创新创业实践平台建设研究”(项目编号:B/2022/01/169);教育部产学合作协同育人项目“中医药高校大数据与人工智能实践教育中心”(项目编号:221005875123454)研究成果之一。

摘　　要：[目的/意义]党史知识学习是当前大众思政学习的重要内容之一,然而大众通过搜索引擎、媒体新闻、公众号等方式获取到的党史知识零碎庞杂且不准确。构建党史知识智能问答系统对宣传和学习百年党史具有一定的应用价值和意义。[方法/过程]对中共中央党校发布的中国共产党一百年大事记数据进行结构化处理,在Neo4j中存储并绘制百年党史知识图谱,以此为基础构建百年党史知识智能问答系统。利用深度学习朴素贝叶斯分类器Naive Bayes、Jieba分词技术和Hanlp分词技术对用户自然问句进行问句分类、问句解析和查询语句转换;使用Ajax技术和Flask框架搭建系统并在局域网内发布。[结果/结论]构建了党史知识智能问答系统,实现了百年党史知识的智能问答功能且准确率高。[Purpose/significance]The study of party history knowledge is one of the important contents of the current public ideological and political learning.But the party history knowledge obtained by the public through search engines,media news,public accounts,etc.,is fragmentary,complex and inaccurate.The construction of the intelligent Q&A system of party history knowledge is of certain application value and significance for publicizing and learning the 100 years of party history.[Method/process]It structures the data of one hundred years of the Communist Party of China published by the Party School of the Central Committee of the Communist Party of China(CPC Central Committee),stores and draws the knowledge map of one hundred years of party history in Neo4j,and constructs an intelligent Q&A system of one hundred years of party history knowledge on the basis of it.It utilizes the deep learning Naive Bayes classifier,the Jieba participle technology,and the Hanlp participle technology to realize the question classification,question parsing and query statement conversion of user’s natural questions,and finally uses Ajax technology and Flask framework to build the system and release it on the LAN.[Result/conclusion]The constructed intelligent Q&A system for party history knowledge can realize the intelligent question and answer function of 100 years of party history knowledge with high accuracy.

关 键 词：百年党史 知识图谱 智能问答系统 深度学习 分词技术 

分 类 号：K335.7[历史地理—历史学]