| 赵浩宇,陈登建,曾桢,张虹雨.基于知识图谱的中国近代史知识问答系统构建研究[J].数字图书馆论坛,2022,(6):31~38 |
| 基于知识图谱的中国近代史知识问答系统构建研究 |
| Research on the Construction of Knowledge Q&A System of Modern Chinese History Based on Knowledge Graph |
| 投稿时间：2022-06-04 |
| DOI：10.3772/j.issn.1673-2286.2022.06.005 |
| 中文关键词: 问答系统；知识图谱；领域本体；中国近代史 |
| 英文关键词: Q&A System; Knowledge Graph; Domain Ontology; Modern Chinese History |
| 基金项目:本研究得到国家自然科学基金项目“基于知识图谱的农产品价值链信息融合研究”（编号：2020XSXM）资助。 |
| 作者 | 单位 | | 赵浩宇 | 贵州财经大学信息学院 | | 陈登建 | 贵州财经大学信息学院 | | 曾桢 | 贵州财经大学信息学院 | | 张虹雨 | 贵州财经大学信息学院 |
|
| 摘要点击次数: 1273 |
| 全文下载次数: 1321 |
| 中文摘要: |
| 历史文献内容通常以线性文本的形式供用户浏览阅读，但当用户意在获取非线性的碎片化信息时，由于普通搜索引擎缺乏理解用户检索意图，且传统的信息组织方式也无法对历史知识元素进行细粒度的语义关联，因而无法完成意图理解式的智能搜索，故本文拟搭建基于知识图谱的中国近代史知识问答系统，以打破传统信息获取方式的局限性。首先从多个维度搭建中国近代史本体模型，并通过知识获取、知识抽取、知识融合、知识存储等方式完成知识图谱的实例构建，然后在此基础上遵循MVC模式三层架构设计实现了一个基于知识图谱的前后端分离的中国近代史知识问答系统，最后优化了知识问答系统中的语句解析任务，并对问答系统的性能进行测试，测试结果取得较好效果，表明该系统能够较为准确地回复自然语言提问。 |
| 英文摘要: |
| The content of historical documents is usually in the form of linear text for users to browse and read. However, when users intend to obtain non-linear fragmented information, the general search engine lacks an understanding of users’ retrieval intention, and the traditional information organization method cannot make fine-grained semantic associations with historical knowledge elements, so it is impossible to complete the intelligent search with intention understanding. Therefore, this paper plans to build a knowledge question-and-answer system of modern Chinese history based on a knowledge graph to break the limitations of traditional information acquisition methods. In this paper, we build an ontology model of modern Chinese history from multiple dimensions and complete the instance construction of a knowledge graph through knowledge acquisition, knowledge extraction, knowledge fusion, knowledge storage, etc. On this basis, we follow the three-layer architecture design of the MVC pattern to realize a knowledge question and answer system based on a knowledge graph with front and back-end separation. Finally, the project optimized the utterance parsing task in the knowledge Q&A system and conducted experimental tests on the performance of the Q&A system, which achieved good results and showed that the system can respond to natural language questions more accurately. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
