该文提出一种新的查询图生成方法用于知识图谱问答系统的问句解析。现有查询图生成工作覆盖的复杂问句类型有限,不能较好地处理答案为关系或涉及关系约束的问句,且未充分考虑路径结果间的组合与运算。因此,该文在查询图生成中应用节点操作的同时引入基于关系的操作,并考虑不同主路径之间的组合情况,显著提升对复杂问句的分析能力。并在此基础上,构建了中文知识图谱问答系统。此外,该文构建一份包含多种复杂类型问句的中文知识图谱问答数据集。该数据集和CCKS2019-CKBQA数据集合并后构成一个新的数据集CCKS2019-Comp,并用来测试本文方法的有效性。实验结果表明,该文方法在CCKS2019-CKBQA和CCKS2019-Comp测试集上平均F1值分别达到73.8%和73.3%。该文的新构建数据和代码已开源①。
Abstract
We propose a new query graph generation method for question parsing in knowledge base question answering system. With a limited coverage of complex questions, existing query graph generation methods fail in handling questions whose answers or constrains are relations or the combination or operation between path results. We employ relation-based actions as well as node-based actions during query graph generation stage, and take the combination of different main paths into consideration. Based on this method, we build a Chinese knowledge base question answering system. We also build a dataset of Chinese knowledge base question answering with multiple complex questions, which is merged with CCKS2019-CKBQA as a new dataset called CCKS2019-Comp. The experimental results show that the proposed method achieves average F1 value of 73.8% and 73.3% on CCKS2019-CKBQA and CCKS2019-Comp, respectively. (data and code are available on GitHub1).
关键词
知识图谱问答 /
查询图生成 /
数据构建 /
问答系统
{{custom_keyword}} /
Key words
knowledge base question answering /
query graph generation /
data construction /
question answering system
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] XU K, LAI Y X, FENG Y S, et al. Enhancing key-value memory neural networks for knowledge based question answering[C]//Proceedings of NAACL, 2019: 2937-2947.
[2] SUN H T, TANIA B W, WILLIAM C. PullNet: Open domain question answering with iterative retrieval on knowledge bases and text[C]//Proceedings of EMNLP, 2019: 2380-2390.
[3] APOORV S, ADITAY T, PARTHA T. Improving multi-hop question answering over knowledge graphs using knowledge base embeddings[C]//Proceedings of ACL, 2020: 4498-4507.
[4] NIKITA B, ZHENG X Y, QIAN K, et al. Answering complex questions by combining information from curated and extracted knowledge bases[C]//Proceedings of ACL, 2020: 1-10.
[5] BERANT J, CHOU A, FROSTIG R, et al. Semantic parsing on free base from question-answer pairs[C]//Proceedings of EMNLP, 2013: 1533-1544.
[6] SUN Y W, ZHANG L L, CHENG C, et al. SPARQA: Skeleton-based semantic parsing for complex questions over knowledge bases[C]//Proceedings of AAAI, 2020, 34(05): 8952-8959.
[7] LAN Y S, JIANG J. Query graph generation for answering multi-hop complex questions from knowledge bases[C]//Proceedings of ACL, 2020: 969-974.
[8] YIH W T, CHANG M W, HE X D, et al. Semantic parsing via staged query graph generation: Question answering with knowledge base[C]//Proceedings of ACL, 2015: 1321-1331.
[9] LUO K Q, LIN F L, LUO X S, et al. Knowledge base question answering via encoding of complex query graphs[C]//Proceedings of EMNLP, 2018: 2185-2194.
[10] PAVAN K, IBRAHIM A, SRINIVAS R, et al. Leveraging abstract meaning representation for knowledge base question answering[C]//Proceedings of ACL, 2021: 3884-3894.
[11] QIN K C, LI C, VIRGIL P, et al. Improving query graph generation for complex question answering over knowledge base[C]//Proceedings of EMNLP, 2021: 4201-4207.
[12] MAHESHWARI G, TRIVEDI P, LUKOVNIKOV D, et al. Learning to rank query graphs for complex question answering over knowledge graphs[C]//Proceedings of ISWC, 2019: 487-504.
[13] 宋鹏程,单丽莉,孙承杰, 等.基于查询路径排序的知识库问答系统[J].中文信息学报,2021,35(11):109-126.
[14] 孙亚伟,程龚,厉肖,等.基于图匹配网络的可解释知识图谱复杂问答方法[J].计算机研究与发展,2021,58(12):2673-2683.
[15] DAI W, LIU H W, LIU Y, et al.一种基于路径挖掘、匹配和排序的知识库问答方法[EB/OL]. http://sigkg.cn/ccks2021/?page_id=669[2022-05-31].
[16] 张鹏举,贾永辉,陈文亮.基于多特征实体消歧的中文知识图谱问答[J].计算机工程,2022,48(2):47-54.
[17] CHEN S, LIU Q, YU Z W, et al. ReTraCk: A flexible and efficient framework for knowledge base question answering[C]//Proceedings of ACL: System Demonstrations, 2021: 325-336.
[18] JIA Y H, CHEN W L. Better query graph selection for knowledge base question answering[EB/OL]. https://arxiv.org/abs/2204.12662[2022-05-31].
[19] 骆金昌, 尹存祥, 吴晓晖, 等. 混合语义相似度的中文知识图谱问答系统[EB/OL]https://conference.bj.bcebos.com/ccks2019/eval/webpage/index.html[2022-05-31].
[20] ZHANG P J, WU K, ZHU Z K, et al.Combining neural network models with rules for chinese knowledge base question answering[EB/OL]https://conference.bj.bcebos.com/ccks2019/eval/webpage/index.html[2022-05-31].
[21] YANG Y Y, HE X H, ZHOU K J, et al. Multi-module system for open domain Chinese question answering over knowledge base[EB/OL]https://conference.bj.bcebos.com/ccks2019/eval/webpage/index.html[2022-05-31].
[22] 吴锟,周夏冰,李正华, 等.中文知识库问答中的路径选择[J].中文信息学报,2021,35(09):113-122.
[23] 王鑫雷,李帅驰,杨志豪, 等.基于预训练语言模型的中文知识图谱问答系统[J].山西大学学报(自然科学版),2020,43(04):955-962.
[24] 彭怀,宋井宽,唐向红.基于信息匹配方法的中文知识库问答系统[J].计算机技术与发展,2022,32(02):14-19.
[25] ZHANG M H, ZHANG R Y, ZOU L, et al. NAMER: A node-based multitasking framework for multi-hop knowledge base question answering[C]//Proceedings of NAACL, 2021: 18-25.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国家自然科学基金(61936010);江苏高校优势学科建设工程
{{custom_fund}}
PDF(1544 KB)
