基于层级池化序列匹配的知识图谱复杂问答最优查询图选择方法

doi:10.12305/j.issn.1001-506X.2024.08.16

摘要/Abstract

摘要：

在处理知识图谱复杂问答任务时, 传统的查询图语义解析方法需要在排序阶段对大量结构复杂的候选查询图进行语义编码, 用以获得各自多维特征表示。然而, 在编码过程中采用的全局最大或平均池化操作通常存在对代表性特征提取能力不足的问题。针对以上问题, 提出一种基于层级池化序列匹配的最优查询图选择方法。在实现候选查询图的交互建模过程中, 同时采用层级池化滑动窗口技术分层提取问句和查询图序列对的局部显著性特征与全局语义特征, 使得到的特征向量更好地用于候选查询图的语义匹配打分。所提方法在两个流行的复杂问答数据集MetaQA和WebQuestionsSP上开展广泛实验。实验结果表明: 引入层级池化操作能够有效提取复杂查询图序列的代表性语义特征, 增强原有排序模型的交互编码能力, 有助于进一步提升知识图谱复杂问答系统的性能。

关键词: 知识图谱复杂问答, 查询图语义解析, 层级池化, 交互编码

Abstract:

When dealing with complex question answering task over knowledge graph, traditional semantic parsing method for query graphs requires encoding massive candidate query graphs with complex structures in the ranking stage to obtain their respective multi-dimensional feature representations. However, the global maximum or average pooling operation used during the encoding process often suffers from insufficient extracting capability for representative feature. To address the aforementioned problem, an optimal selection method for query graphs based on hierarchical pooling sequence matching is proposed. Meanwhile, sliding window technique based on hierarchical pooling is adopted to hierarchically extract local salient features and global semantic features of question and query graph sequence pairs during the interactive modeling of candidate query graphs, making the resulting feature vectors better used for semantic matching scoring of candidate query graphs. The proposed method is extensively evaluated on two popular complex question answering datasets, MetaQA and WebQuestionsSP. Experiment results show that by introducing hierarchical pooling operation, representative semantic features of complex query graph sequences can be effectively extracted, and the interactive encoding capability of the original ranking model can be enhanced, which helps further improve the performance of complex question answering systems over knowledge graph.

Key words: complex question answering over knowledge graph, semantic parsing for query graph, hierarchical pooling, interactive encoding

中图分类号:

TP391

王冬, 周思航, 黄健, 张中杰. 基于层级池化序列匹配的知识图谱复杂问答最优查询图选择方法[J]. 系统工程与电子技术, 2024, 46(8): 2686-2695.

Dong WANG, Sihang ZHOU, Jian HUANG, Zhongjie ZHANG. Hierarchical pooling sequence matching based optimal selection method of query graph for complex question answering over knowledge graph[J]. Systems Engineering and Electronics, 2024, 46(8): 2686-2695.

图/表 14

图1

表1

图2

图3

图4

图5

图6

图7

图8

表2

表3

图9

图10

图11

参考文献 42

1	ZHANG J , CHEN B , ZHANG L X , et al. Neural, symbolic and neural-symbolic reasoning on knowledge graphs[J]. AI Open, 2021, 2, 14- 35. doi: 10.1016/j.aiopen.2021.03.001
2	王智悦, 于清, 王楠, 等. 基于知识图谱的智能问答研究综述[J]. 计算机工程与应用, 2020, 56 (23): 1- 11. doi: 10.3778/j.issn.1002-8331.2004-0370
	WANG Z Y , YU Q , WANG N , et al. Survey of intelligent question answering research based on knowledge graph[J]. Computer Engineering and Applications, 2020, 56 (23): 1- 11. doi: 10.3778/j.issn.1002-8331.2004-0370
3	萨日娜, 李艳玲, 林民. 知识图谱推理问答研究综述[J]. 计算机科学与探索, 2022, 16 (8): 1727- 1741.
	SA R N , LI Y L , LIN M . A Survey of question answering based on knowledge graph reasoning[J]. Computer Science and Technology, 2022, 16 (8): 1727- 1741.
4	刘胜. 联合逻辑形式解析和答案生成的知识图谱问答方法及应用[D]. 杭州: 浙江大学, 2023.
	LIU S. Knowledge graph question answering method and application based on logical form parsing and answer generation[D]. Hangzhou: Zhejiang University, 2023.
5	CAO X , ZHAO Y S , SHEN B . Improving and evaluating complex question answering over knowledge bases by constructing strongly supervised data[J]. Neural Computing and Applications, 2023, 35 (7): 5513- 5533. doi: 10.1007/s00521-022-07965-0
6	LAN Y S, WANG S H, JIANG J. Knowledge base question answering with topic units[C]//Proc. of the 28th International Joint Conference on Artificial Intelligence, 2019: 5046-5052.
7	LAN Y S , WANG S H , JIANG J . Knowledge base question answering with a matching-aggregation model and question-specific contextual relations[J]. IEEE/ACM Trans.on Audio, Speech, and Language Processing, 2019, 27 (10): 1629- 1638. doi: 10.1109/TASLP.2019.2926125
8	LAN Y S, HE G, JIANG J, et al. A survey on complex know-ledge base question answering: methods, challenges and solutions[C]//Proc. of the 30th International Joint Conference on Artificial Intelligence, 2021: 4483-4491.
9	XU K, LAI Y X, FENG Y S, et al. Enhancing key-value memory neural networks for knowledge based question answering[C]// Proc. of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019: 2937-2947.
10	SUN H, BEDRAX-WEISS T, COHEN W W. PullNet: open domain question answering with iterative retrieval on know-ledge bases and text[C]//Proc. of the Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, 2019: 2380-2390.
11	SUN H, DHINGRA B, ZAHEER M, et al. Open domain question answering using early fusion of knowledge bases and text[C]//Proc. of the Conference on Empirical Methods in Na-tural Language Processing, 2018: 4231-4242.
12	SAXENA A, TRIPATHI A, TALUKDAR P. Improving multi-hop question answering over knowledge graphs using knowledge base embeddings[C]//Proc. of 58th Annual Meeting of the Association for Computational Linguistics, 2020: 4498-4507.
13	SHI J X, CAO S L, HOU L, et al. TransferNet: an effective and transparent framework for multi-hop question answering over relation graph[C]//Proc. of the Conference on Empirical Methods in Natural Language Processing, 2021: 4149-4158.
14	YIH S W, CHANG M W, HE X, et al. Semantic parsing via staged query graph generation: question answering with know-ledge base[C]//Proc. of the 53rd Annual Meeting of the ACL and the 7th International Joint Conference on Natural Language Processing, 2015: 1321-1331.
15	BAO J W, DUAN N, YAN Z, et al. Constraint-based question answering with knowledge graph[C]//Proc. of the 26th International Conference on Computational Linguistic: Technical Papers, 2016: 2503-2514.
16	MAHESHWARI G, TRIVED P, LUKOVNIKOV D, et al. Learning to rank query graphs for complex question answering over knowledge graphs[C]//Proc. of the 18th International Semantic Web Conference, 2019: 26-30.
17	QIN K, LI C, PAVLU V, et al. Improving query graph generation for complex question answering over knowledge base[C]// Proc. of the Conference on Empirical Methods in Natural Language Processing, 2021: 4201-4207.
18	LAN Y S, JIANG J. Query graph generation for answering multi-hop complex questions from knowledge bases[C]//Proc. of the 58th Annual Meeting of the Association for Computational Linguistics, 2020: 969-974.
19	CHEN Y R, LI H Y, HUA Y C, et al. Formal query building with query structure prediction for complex question answering over knowledge base[C]//Proc. of the 29th International Joint Conference on Artificial Intelligence, 2021: 3751-3758.
20	LI M, JI S H. Semantic structure based query graph prediction for question answering over knowledge graph[C]//Proc. of the 29th International Conference on Computational Linguistics, 2022: 1569-1579.
21	LUO K Q, LIN F L, LUO X S, et al. Knowledge base question answering via encoding of complex query graphs[C]//Proc. of the Conference on Empirical Methods in Natural Language Processing, 2018: 2185-2194.
22	JIA Y H, CHEN W L. Better query graph selection for know-ledge base question answering[EB/OL]. [2022-10-22]. https://doi.org/10.48550/arXiv.2204.12662.
23	DEVLIN J, CHANG M W, LEE K, et al. BERT: pre-training of deep bidirectional transformers for language understanding[C]// Proc. of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019: 4171-4186.
24	YIN X , GROMANN D , RUDOLPH S . Neural machine translating from natural language to SPARQL[J]. Future Generation Computer Systems, 2021, 117, 510- 519. doi: 10.1016/j.future.2020.12.013
25	马玉凤, 向南, 豆亚杰, 等. 军事系统工程中的知识图谱应用及研究[J]. 系统工程与电技术, 2022, 44 (1): 146- 153.
	MA Y F , XIANG N , DOU Y J , et al. Application and research of knowledge graph in military system engineering[J]. Systems Engineering and Electronics, 2022, 44 (1): 146- 153.
26	封皓君, 段立, 张碧莹, 等. 双向循环进化的实体链接及知识推理框架[J]. 系统工程与电技术, 2022, 44 (9): 2878- 2885.
	FENG H J , DUAN L , ZHANG B Y , et al. Bidirectional cyclic evolutionary framework of entity linking and knowledge reasoning[J]. Systems Engineering and Electronics, 2022, 44 (9): 2878- 2885.
27	HUA Y, LI Y F, HAFFARI G, et al. Few-shot complex know-ledge base question answering via meta reinforcement learning[C]// Proc. of the Conference on Empirical Methods in Natural Language Processing, 2020: 5827-5837.
28	HUA Y , LI Y F , QI G , et al. Less is more: data-efficient complex question answering over knowledge bases[J]. Journal of Web Semantics, 2020, 65, 100612. doi: 10.1016/j.websem.2020.100612
29	ZHANG Q X , WENG X Y , ZHOU G Y , et al. RL: an adaptive reinforcement learning framework for complex question answering over knowledge base[J]. Information Processing and Management, 2022, 59 (3): 102933. doi: 10.1016/j.ipm.2022.102933
30	BAST H, HAUSSMANN E. More accurate question answering on freebase[C]//Proc. of the 24th ACM International on Conference on Information and Knowledge Management, 2015: 1431-1440.
31	JIA Z, ABUJABAL A, SAHA R R, et al. Temporal question answering over knowledge bases[C]// Proc. of the 27th ACM International Conference on Information and Knowledge Management, 2018: 1807-1810.
32	SUN Z, DENG Z H, NIE J Y, et al. Rotate: knowledge graph embedding by relational rotation in complex space[EB/OL]. [2022-10-22]. https://doi.org/10.48550/arXiv.1902.10197.
33	杨东华, 邹开发, 王宏志, 等. 基于Seq2Seq模型的SparQL查询预测[J]. 软件学报, 2021, 32 (3): 805- 817.
	YANG D H , ZOU K F , WANG H Z , et al. SparQL query prediction based on Seq2Seq model[J]. Journal of Software, 2021, 32 (3): 805- 817.
34	KRIZHEVSKY A , SUTSKEVER I , HINTON G E . Imagenet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60 (6): 84- 90. doi: 10.1145/3065386
35	SNELL J, SWERSKY K, ZEMEL R, et al. Prototypical networks for few-shot learning[C]//Proc. of the 31st International Conference on Neural Information Processing Systems, 2017: 4080-4090.
36	SHEN D H, WANG G Y, WANG W L, et al. Baseline needs more love: on simple word-embedding-based models and associated pooling mechanisms[C]//Proc. of the 56th Annual Meeting of the Association for Computational Linguistics, 2018: 440-450.
37	PAN C Y , HUANG J , GONG J X . Few-shot learning with hierarchical pooling induction network[J]. Multimedia Tools and Applications, 2022, 81 (23): 32937- 32952. doi: 10.1007/s11042-022-11999-w
38	SCHROFF F, KALENICHENKO D, PHILBIN J. Facenet: a unified embedding for face recognition and clustering[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 815-823.
39	ZHANG Y, DAI H, KOZAREVA Z, et al. Variational reasoning for question answering with knowledge graph[C]// Proc. of the 32nd AAAI Conference on Artificial Intelligence, 2018.
40	YIH W, RICHARDSON M, MEEK C, et al. The value of semantic parse labeling for knowledge base question answering[C]// Proc. of the 54th Annual Meeting of the Association for Computational Linguistics, 2016: 201-206.
41	CHEN Z Y, CHANG C H, CHEN Y P, et al. UHop: an unrestricted-hop relation extraction framework for knowledge-based question answering[C]//Proc. of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019: 345-356.
42	HE G L, LAN Y S, JIANG J, et al. Improving multi-hop knowledge base question answering by learning intermediate supervision signals[C]//Proc. of the 14th ACM International Conference on Web Search and Data Mining, 2021: 553-561.

查询图组成	类别	实例
顶点集合	知识图谱实体	特朗普, 北京, …
	知识图谱类型	电子, 演员, 飞行, …
	数字	1, 2022, …
	中间变量或答案变量	V₀, V₁, …
边集合	知识图谱关系	< 工作于>, < 居住在>, …
	次序	min_at_n, max_at_n, …
	比较	< , >, =, …
	统计	+, -, …
	⋮	⋮

数据集	训练集	验证集	测试集
MetaQA_1-hop	96 106	7 192	7 147
MetaQA_2-hop	118 980	14 872	14 872
MetaQA_3-hop	114 196	14 274	14 274
WebQuestionsSP	3 304	—	1 815

模型	MetaQA			WebQuestionsSP
模型	1-hop	2-hop	3-hop	WebQuestionsSP
PullNet	97.0	99.9	91.4	68.1
Uhop	—	—	—	67.2
QGG	—	—	—	71.7
EmbedKGQA	97.5	98.8	94.8	66.6
TransferNet	96.0	98.5	94.7	71.4
NSM	97.1	99.9	98.9	68.7
SSKGQA	99.1	99.7	99.6	71.4
所提方法	99.3	99.8	99.8	71.7

[1]	费博雯, 包卫东, 刘大千, 朱晓敏. 面向动态目标搜索与打击的空地协同自主任务分配方法[J]. 系统工程与电子技术, 2024, 46(7): 2346-2358.
[2]	余婧, 吴晓军, 蒋安林, 雍恩米. 基于多精度规划窗口的无人机航迹规划方法研究[J]. 系统工程与电子技术, 2024, 46(5): 1767-1776.
[3]	罗俊仁, 张万鹏, 苏炯铭, 袁唯淋, 陈璟. 多智能体博弈学习研究进展[J]. 系统工程与电子技术, 2024, 46(5): 1628-1655.
[4]	吴钇达, 王彩云, 王佳宁, 李晓飞. 基于ISVM-DS的红外多传感器融合识别方法[J]. 系统工程与电子技术, 2024, 46(5): 1555-1560.
[5]	黄思佳, 宋纯锋, 李璇. 基于可变尺度先验框的声呐图像目标检测[J]. 系统工程与电子技术, 2024, 46(3): 771-778.
[6]	马兰, 孟诗君, 吴志军. 基于BERT与生成对抗的民航陆空通话意图挖掘[J]. 系统工程与电子技术, 2024, 46(2): 740-750.
[7]	胡涛, 申立群, 朱镜达, 孙成会, 董伟锋. 基于FAST和Sobol指数法的雷达系统效能敏感性分析[J]. 系统工程与电子技术, 2024, 46(2): 561-569.
[8]	曲熠, 陈莹. 基于边缘强化的无监督单目深度估计[J]. 系统工程与电子技术, 2024, 46(1): 71-79.
[9]	颜骥, 刘丙杰, 陈建华. 基于电子商务竞标结构的分布式作战资源调度[J]. 系统工程与电子技术, 2024, 46(1): 227-236.
[10]	张宝琛, 惠建江, 张琦, 刘正雄, 黄攀峰. 面向冗余机械臂避障运动规划的触控交互技术[J]. 系统工程与电子技术, 2024, 46(1): 254-260.
[11]	黄万炎, 杜万和, 杨淑珍, 俞涛. 改进的多项式曲线拟合轨迹预测算法[J]. 系统工程与电子技术, 2024, 46(1): 280-289.
[12]	贺翥祯, 李敏, 苟瑶, 杨爱涛. 改进YOLOv5的合成孔径雷达图像舰船目标检测方法[J]. 系统工程与电子技术, 2023, 45(12): 3743-3753.
[13]	胡涛, 申立群, 田宇阳, 董伟锋. 航天复杂系统测发控流程仿真引擎设计与评价[J]. 系统工程与电子技术, 2023, 45(12): 3866-3874.
[14]	翟一琛, 顾佼佼, 宗富强, 姜文志. 融合注意力机制的IETM细粒度跨模态检索算法[J]. 系统工程与电子技术, 2023, 45(12): 3915-3923.
[15]	陈维义, 何凡, 刘国强, 毛伟伟. 变结构交互式多模型滤波和平滑算法[J]. 系统工程与电子技术, 2023, 45(12): 4005-4012.