您的位置：首页 > 论文页面
知识图谱在问答系统中的应用综述
发表时间：2017-05-31 浏览量：5218 下载量：1441
| 全部作者： | 刘俪婷，师文轩，程书芝 |
| 作者单位： | 南开大学软件学院 |
| 摘 要： | 随着互联网文档万维网向数据万维网的转型，各大搜索引擎公司纷纷构建知识图谱。为更好地理解知识图谱的内涵及其利用价值，对相关领域的研究进行汇总和梳理。首先，介绍问答系统的起源发展和知识图谱的概论。然后，分别从概念层次关系学习、命名实体识别与实体链接、事实知识学习和事件知识学习四个方面的信息抽取技术介绍知识图谱构建技术。最后，将知识图谱应用到问答系统中，得出知识图谱的构建有望有效地扩大知识库的规模、给问答系统的进一步研究提供新方向的结论。然而，知识图谱的发展机遇与挑战并存，仍需进一步研究与积累。 |
| 关 键 词： | 计算机应用；知识图谱；综述；问答系统；信息抽取；深度学习 |
| Title： | Review of the application of knowledge graph in question answering system |
| Author： | LIU Liting, SHI Wenxuan, CHENG Shuzhi |
| Organization： | College of Software, Nankai University |
| Abstract： | With the transition of the Internet from the world wide document web to the world wide data web, major search engine companies turn to build the knowledge graph. This paper summarizes the researches in the field of knowledge graph in order to make a better understanding of the connotation of knowledge graph and its utilization value. Firstly, the origin and development of question answering system and the generality of knowledge graph are introduced. Then, the techniques of building knowledge graph by information extraction techniques are presented. Information extraction techniques are divided into four aspects, including concept hierarchical relationship learning, named entity recognition and entity linking, factual knowledge learning and event knowledge learning. Finally, knowledge graph is applied to question answering system, and it can be concluded that the knowledge graph is expected to expand the scale of knowledge base effectively and provide a new orientation for the further study of question answering system. However, the development of knowledge graph has both opportunities and challenges, which calls for further researches to gain a deeper understanding on knowledge graph. |
| Key words： | computer application; knowledge graph; review; question answering system; information extraction; deep learning |
| 发表期数： | 2017年5月第10期 |
| 引用格式： | 刘俪婷，师文轩，程书芝. 知识图谱在问答系统中的应用综述[J]. 中国科技论文在线精品论文，2017，10（10）：1052-1067. |
2
评论数 0请您登录
暂无评论
下载全文
多维论文