《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (4): 1065-1071.DOI: 10.11772/j.issn.1001-9081.2021071265
所属专题: CCF第36届中国计算机应用大会 (CCF NCCA 2021)
• CCF第36届中国计算机应用大会 (CCF NCCA 2021) • 上一篇 下一篇
余晓鹏1,2, 何儒汉1,2, 黄晋2,3(), 张俊杰1,2, 胡新荣2,3
收稿日期:
2021-07-16
修回日期:
2021-08-20
接受日期:
2021-08-25
发布日期:
2021-08-20
出版日期:
2022-04-10
通讯作者:
黄晋
作者简介:
余晓鹏(1994—),男,山西大同人,硕士研究生,CCF会员,主要研究方向:知识图谱、自然语言处理基金资助:
Xiaopeng YU1,2, Ruhan HE1,2, Jin HUANG2,3(), Junjie ZHANG1,2, Xinrong HU2,3
Received:
2021-07-16
Revised:
2021-08-20
Accepted:
2021-08-25
Online:
2021-08-20
Published:
2022-04-10
Contact:
Jin HUANG
About author:
YU Xiaopeng, born in 1994, M. S. candidate. His research interests include knowledge graph, natural language processing.Supported by:
摘要:
知识图谱嵌入(KGE)将实体和关系映射到低维连续向量空间中,以利用机器学习方法实现关系数据的应用,如知识分析、推理、补全等。以ConvE为代表将卷积神经网络(CNN)应用于知识图谱嵌入中,以捕捉实体和关系的交互信息,但其标准卷积捕捉特征交互信息能力不足,特征表达能力低下。针对特征交互能力不足问题,提出了一种改进的Inception结构,在此基础上构建一个知识图谱嵌入模型InceE。首先,该结构使用混合空洞卷积替代标准卷积,以提高特征交互信息捕捉能力;其次,使用残差网络结构,以减少特征信息丢失。实验使用基准数据集Kinship、FB15k、WN18验证InceE链接预测有效性。在Kinship、FB15k数据集上,相较于ArcE和QuatRE模型,InceE的Hit@1分别提升了1.6和1.5个百分点;在三个数据集上,与ConvE对比,InceE的Hit@1分别提升了6.3、20.8和1.0个百分点。实验结果表明InceE具有更强的特征交互信息捕捉能力。
中图分类号:
余晓鹏, 何儒汉, 黄晋, 张俊杰, 胡新荣. 基于改进Inception结构的知识图谱嵌入模型[J]. 计算机应用, 2022, 42(4): 1065-1071.
Xiaopeng YU, Ruhan HE, Jin HUANG, Junjie ZHANG, Xinrong HU. Knowledge graph embedding model based on improved Inception structure[J]. Journal of Computer Applications, 2022, 42(4): 1065-1071.
模型 | 评分函数 | 参数 | 模型 | 评分函数 | 参数 |
---|---|---|---|---|---|
TransE | DisMult | ||||
TransR | CompIEx | ||||
ManifoldE | ConvE | ||||
RotatE | InceE | ||||
Rescal |
表1 知识图谱嵌入模型及评分函数
Tab. 1 Knowledge graph embedding model and scoring function
模型 | 评分函数 | 参数 | 模型 | 评分函数 | 参数 |
---|---|---|---|---|---|
TransE | DisMult | ||||
TransR | CompIEx | ||||
ManifoldE | ConvE | ||||
RotatE | InceE | ||||
Rescal |
数据集 | #Relation | #Entity | #Triplet | ||
---|---|---|---|---|---|
Train | Valid | Test | |||
FB15k | 1 345 | 14 951 | 483 142 | 50 000 | 59 071 |
WN18 | 18 | 40 943 | 141 442 | 5 000 | 5 000 |
Kinship | 25 | 104 | 8 544 | 1 068 | 1 074 |
表2 数据集数据统计
Tab. 2 Dataset statistics
数据集 | #Relation | #Entity | #Triplet | ||
---|---|---|---|---|---|
Train | Valid | Test | |||
FB15k | 1 345 | 14 951 | 483 142 | 50 000 | 59 071 |
WN18 | 18 | 40 943 | 141 442 | 5 000 | 5 000 |
Kinship | 25 | 104 | 8 544 | 1 068 | 1 074 |
模型 | MRR | Hit@1/% | Hit@3/% | Hit@10/% |
---|---|---|---|---|
CompIEx | 0.823 | 73.3 | 89.9 | 97.1 |
ConvE | 0.833 | 73.8 | 91.7 | 98.1 |
ConvKB | 0.614 | 43.6 | 75.5 | 95.3 |
R-GCN | 0.109 | 3.0 | 8.8 | 23.9 |
SimplE | 0.752 | 62.6 | 85.4 | 97.2 |
RotatE | 0.843 | 76.0 | 91.9 | 97.8 |
HAKE | 0.852 | 76.9 | 92.8 | 98.0 |
InteractE | 0.777 | 66.4 | 87.0 | 95.9 |
CompGCN | 0.778 | 66.7 | 86.8 | 96.7 |
CoKE | 0.793 | 69.3 | 87.8 | 95.4 |
ArcE | 0.864 | 78.5 | 93.9 | 98.7 |
InceE | 0.873 | 80.1 | 93.7 | 98.3 |
表3 不同模型在Kinship数据集的实验结果
Tab. 3 Experimental results of different models on Kinship dataset
模型 | MRR | Hit@1/% | Hit@3/% | Hit@10/% |
---|---|---|---|---|
CompIEx | 0.823 | 73.3 | 89.9 | 97.1 |
ConvE | 0.833 | 73.8 | 91.7 | 98.1 |
ConvKB | 0.614 | 43.6 | 75.5 | 95.3 |
R-GCN | 0.109 | 3.0 | 8.8 | 23.9 |
SimplE | 0.752 | 62.6 | 85.4 | 97.2 |
RotatE | 0.843 | 76.0 | 91.9 | 97.8 |
HAKE | 0.852 | 76.9 | 92.8 | 98.0 |
InteractE | 0.777 | 66.4 | 87.0 | 95.9 |
CompGCN | 0.778 | 66.7 | 86.8 | 96.7 |
CoKE | 0.793 | 69.3 | 87.8 | 95.4 |
ArcE | 0.864 | 78.5 | 93.9 | 98.7 |
InceE | 0.873 | 80.1 | 93.7 | 98.3 |
模型 | FB15k | |||
---|---|---|---|---|
MRR | Hit@1/% | Hit@3/% | Hit@10/% | |
TransE | 0.463 | 29.7 | 57.8 | 74.9 |
HOIE | 0.524 | 40.2 | 61.3 | 73.9 |
SimplEx | 0.727 | 66.0 | 77.3 | 83.8 |
ConvE | 0.657 | 55.8 | 72.3 | 83.1 |
R-GCN | 0.696 | 60.1 | 76.0 | 84.2 |
RotatE | 0.797 | 74.6 | 83.0 | 88.4 |
RSNs | 0.780 | 72.2 | ― | 87.3 |
QuatRE | 0.808 | 75.1 | 85.1 | 89.6 |
InceE | 0.815 | 76.6 | 85.0 | 89.6 |
表4 不同模型在FB15k数据集的实验比较结果
Tab. 4 Experimental comparison results of different models on FB15k dataset
模型 | FB15k | |||
---|---|---|---|---|
MRR | Hit@1/% | Hit@3/% | Hit@10/% | |
TransE | 0.463 | 29.7 | 57.8 | 74.9 |
HOIE | 0.524 | 40.2 | 61.3 | 73.9 |
SimplEx | 0.727 | 66.0 | 77.3 | 83.8 |
ConvE | 0.657 | 55.8 | 72.3 | 83.1 |
R-GCN | 0.696 | 60.1 | 76.0 | 84.2 |
RotatE | 0.797 | 74.6 | 83.0 | 88.4 |
RSNs | 0.780 | 72.2 | ― | 87.3 |
QuatRE | 0.808 | 75.1 | 85.1 | 89.6 |
InceE | 0.815 | 76.6 | 85.0 | 89.6 |
模型 | WN18 | |||
---|---|---|---|---|
MRR | Hit@1/% | Hit@3/% | Hit@10/% | |
TransE | 0.495 | 11.3 | 88.8 | 94.3 |
HOIE | 0.938 | 93.0 | 94.5 | 94.9 |
SimplEx | 0.941 | 93.6 | 94.5 | 94.7 |
ConvE | 0.942 | 93.5 | 94.7 | 95.5 |
R-GCN | 0.696 | 60.1 | 76.0 | 84.2 |
RotatE | 0.949 | 94.4 | 95.2 | 95.9 |
RSNs | 0.940 | 92.2 | ― | 95.3 |
QuatRE | 0.939 | 92.3 | 95.3 | 96.3 |
InceE | 0.949 | 94.5 | 95.1 | 95.5 |
表5 不同模型在WN18数据集的实验比较结果
Tab. 5 Experimental comparison results of different models on WN18 dataset
模型 | WN18 | |||
---|---|---|---|---|
MRR | Hit@1/% | Hit@3/% | Hit@10/% | |
TransE | 0.495 | 11.3 | 88.8 | 94.3 |
HOIE | 0.938 | 93.0 | 94.5 | 94.9 |
SimplEx | 0.941 | 93.6 | 94.5 | 94.7 |
ConvE | 0.942 | 93.5 | 94.7 | 95.5 |
R-GCN | 0.696 | 60.1 | 76.0 | 84.2 |
RotatE | 0.949 | 94.4 | 95.2 | 95.9 |
RSNs | 0.940 | 92.2 | ― | 95.3 |
QuatRE | 0.939 | 92.3 | 95.3 | 96.3 |
InceE | 0.949 | 94.5 | 95.1 | 95.5 |
模型 | MRR | Hit@1/% | Hit@3/% | Hit@10/% |
---|---|---|---|---|
ConvE | 0.833 | 73.8 | 91.7 | 98.1 |
InteractE | 0.777 | 66.4 | 87.0 | 95.9 |
ArcE | 0.864 | 78.5 | 93.9 | 98.7 |
InceE | 0.873 | 80.1 | 93.7 | 98.3 |
表6 基于CNN的不同模型在Kinship数据集上的实验结果
Tab. 6 Experimental results of different models based on CNN on Kinship dataset
模型 | MRR | Hit@1/% | Hit@3/% | Hit@10/% |
---|---|---|---|---|
ConvE | 0.833 | 73.8 | 91.7 | 98.1 |
InteractE | 0.777 | 66.4 | 87.0 | 95.9 |
ArcE | 0.864 | 78.5 | 93.9 | 98.7 |
InceE | 0.873 | 80.1 | 93.7 | 98.3 |
模型 | MRR | Hit@1/% | Hit@3/% | Hit@3/% |
---|---|---|---|---|
InceE-Residual | 0.860 | 77.9 | 93.6 | 98.3 |
InceE | 0.873 | 80.1 | 93.7 | 98.3 |
表7 InceE模型是否添加残差学习在Kinship数据集上的实验结果
Tab. 7 Experimental results of InceE model whether to add residual learning on Kinship dataset
模型 | MRR | Hit@1/% | Hit@3/% | Hit@3/% |
---|---|---|---|---|
InceE-Residual | 0.860 | 77.9 | 93.6 | 98.3 |
InceE | 0.873 | 80.1 | 93.7 | 98.3 |
1 | BOLLACKER K, EVANS C, PARITOSH P, et al. Freebase: a collaboratively created graph database for structuring human knowledge[C]//Proceedings of the 2008 ACM SIGMOD International Conference on Management of Data. New York:ACM, 2008: 1247-1250. 10.1145/1376616.1376746 |
2 | MILLER G A. WordNet: a lexical database for English[J]. Communications of the ACM,1995, 38(11): 39-41. 10.1145/219717.219748 |
3 | SUCHANEK F M, KASNECI G, WEIKUM G. YAGO: a core of semantic knowledge[C]// Proceedings of the 16th International Conference on World Wide Web. New York: ACM, 2007: 697-706. 10.1145/1242572.1242667 |
4 | ZHANG F Z, YUAN N J, LIAN D, et al. Collaborative knowledge base embedding for recommender systems[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2016: 353-362. 10.1145/2939672.2939673 |
5 | 刘知远,孙茂松,林衍凯,等.知识表示学习研究进展[J].计算机研究与发展,2016,53(2):247-261. 10.7544/issn1000-1239.2016.20160020 |
LIU Z Y, SUN M S, LIN Y K, et al. Knowledge representation learning: a review[J].Journal of Computer Research and Development,2016,53(2):247-261. 10.7544/issn1000-1239.2016.20160020 | |
6 | BORDES A, USUNIER N, GARCIA-DURAN A, et al. Translating embeddings for modeling multi-relational data[C]// Proceedings of the 26th International Conference on Neural Information Processing Systems. New York: ACM, 2013: 2787-2795. 10.1007/978-3-662-44848-9_28 |
7 | NICKEL M, TRESP V, KRIEGEL H P. A three-way model for collective learning on multi-relational data[C]// Proceedings of the 28th International Conference on Machine Learning. New York: ACM, 2011: 809-816. 10.1145/2187836.2187874 |
8 | DETTMERS T, MINERVINI P, STENETORP P, et al. Convolutional 2D knowledge graph embeddings[C]// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2018: 1811-1818. |
9 | SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 1-9. 10.1109/cvpr.2015.7298594 |
10 | HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 770-778. 10.1109/cvpr.2016.90 |
11 | WANG Z, ZHANG J W, FENG J L, et al. Knowledge graph embedding by translating on hyperplanes[C]// Proceedings of the 28th AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2014: 1112-1119. 10.1609/aaai.v33i01.33017152 |
12 | LIN Y, LIU Z, SUN M,et al. Learning entity and relation embeddings for knowledge graph completion[C]// Proceedings of the 29th AAAI Conference on Artificial Intelligence. Palo Alto:AAAI Press,2015:2181-2187. |
13 | XIAO H, HUANG M L, ZHU X Y. From one point to a manifold: Knowledge graph embedding for precise link prediction[C]// Proceedings of the 25th International Joint Conference on Artificial Intelligence. Palo Alto:AAAI Press, 2016: 1315-1321. |
14 | SUN Z Q, DENG Z H, NIE J Y, et al. RotatE: Knowledge graph embedding by relational rotation in complex space[C]// Proceedings of the 2019 International Conference on Learning Representations. New Orleans: ICLR, 2019: 1-18. |
15 | KAZEMI S M, POOLE D. Simple embedding for link prediction in knowledge graphs[C]// Proceedings of the 32nd International Conference on Neural Information Processing Systems. Red Hook: Curran Associates, Inc., 2018: 4289-4300. |
16 | YANG B S, YIH W T, HE X D, et al. Embedding entities and relations for learning and inference in knowledge bases[EB/OL]. [2015-08-29]. . |
17 | TROUILLON T, WELBL J, RIEDEL S, et al. Complex embeddings for simple link prediction[C]// Proceedings of the 33rd International Conference on Machine Learning. New York: PMLR.org, 2016:2071-2080. |
18 | LIU H X, WU Y X, YANG Y M. Analogical inference for multi-relational embeddings[C]// Proceedings of the 34th International Conference on Machine Learning. New York: PMLR.org, 2017: 2168-2178. |
19 | DONG X, GABRILOVICH E, HEITZ G, et al. Knowledge vault: a web-scale approach to probabilistic knowledge fusion[C]// Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2014: 601-610. 10.1145/2623330.2623623 |
20 | SOCHER R, CHEN D Q, MANNING C D, et al. Reasoning with neural tensor networks for knowledge base completion[C]// Proceedings of the 26th Neural Information Processing Systems. Red Hook: Curran Associates, Inc., 2013: 926-934. |
21 | NGUYEN D Q, NGUYEN T, NGUYEN D Q, et al. A novel embedding model for knowledge base completion based on convolutional neural network[C]// Proceedings of the 2018 North American Chapter of the Association for Computational Linguistics. Stroudsburg, PA: Association for Computational Linguistics, 2018: 327-333. 10.18653/v1/n18-2053 |
22 | VASHISHTH S, SANYAL S, NITIN V, et al. InteractE: Improving convolution-based knowledge graph embeddings by increasing feature interactions[C]// Proceedings of the 2020 AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2020: 3009-3016. 10.1609/aaai.v34i03.5694 |
23 | SCHLICHTKRULL M, KIPF T N, BLOEM P, et al. Modeling relational data with graph convolutional networks[C]// Proceedings of the 2018 European Semantic Web Conference. Cham: Springer, 2018: 593-607. 10.1007/978-3-319-93417-4_38 |
24 | LIN X V, SOCHER R, XIONG C M. Multi-hop knowledge graph reasoning with reward shaping[C]// Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA: Association for Computational Linguistics, 2018:3243-3253. 10.18653/v1/d18-1362 |
25 | ZHANG Z Q, CAI J Y, ZHANG Y D, et al. Learning hierarchy-aware knowledge graph embeddings for link prediction[C]// Proceedings of the 2020 AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2020: 3065-3072. 10.1609/aaai.v34i03.5701 |
26 | VASHISHTH S, SANYAL S, NITIN V, et al. Composition-based multi-relational graph convolutional networks[C]// Proceedings of the Eighth International Conference on Learning Representations, New Orleans: ICLR, 2020: 1-14. 10.1609/aaai.v34i03.5694 |
27 | WANG Q, HUANG P P, WANG H F, et al. CoKE: Contextualized knowledge graph embedding [EB/OL]. [2020-04-04]. . |
28 | REN F L, LI J C, ZHANG H H, et al. Knowledge graph embedding with atrous convolution and residual learning[C]// Proceedings of the 28th International Conference on Computational Linguistics. Barcelona: International Committee on Computational Linguistics, 2020: 1532-1543. 10.18653/v1/2020.coling-main.134 |
29 | NICKEL M, ROSASCO L, POGGIO T. Holographic embeddings of knowledge graphs[C]// Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2016:1955-1961. |
30 | GUO L B, SUN Z Q, HU W. Learning to exploit long-term relational dependencies in knowledge graphs[C]// Proceedings of the 36th International Conference on Machine Learning. Birmingham: PMLR, 2019: 2505-2514. 10.1162/dint_a_00016 |
31 | NGUYEN D Q, VU T, NGUYEN T D, et al. QuatRE: relation-aware quaternions for knowledge graph embeddings[EB/OL].[2020-09-26]. . 10.1109/tpami.2022.3147798/mm1 |
[1] | 李豆豆, 李汪根, 夏义春, 束阳, 高坤. 基于特征交互与自适应融合的骨骼动作识别[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2581-2587. |
[2] | 王可可, 朱宇, 王晓英, 黄建强, 曹腾飞. 超边约束的异质超网络表示学习方法[J]. 《计算机应用》唯一官方网站, 2023, 43(12): 3654-3661. |
[3] | 袁立宁, 刘钊. 基于One-Shot聚合自编码器的图表示学习[J]. 《计算机应用》唯一官方网站, 2023, 43(1): 8-14. |
[4] | 靳华中, 张修洋, 叶志伟, 张闻其, 夏小鱼. 基于近似U型网络结构的图像去噪模型[J]. 《计算机应用》唯一官方网站, 2022, 42(8): 2571-2577. |
[5] | 陈恒, 王思懿, 李正光, 李冠宇, 刘鑫. 基于关系记忆的胶囊网络知识图谱嵌入模型[J]. 《计算机应用》唯一官方网站, 2022, 42(7): 1985-1992. |
[6] | 韩滕跃, 牛少彰, 张文. 基于对比学习的多模态序列推荐算法[J]. 《计算机应用》唯一官方网站, 2022, 42(6): 1683-1688. |
[7] | 曹一珉, 蔡磊, 高敬阳. 基于生成对抗网络的基因数据生成方法[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 783-790. |
[8] | 潘仁志, 钱付兰, 赵姝, 张燕平. 基于卷积神经网络交互的用户属性偏好建模的推荐模型[J]. 《计算机应用》唯一官方网站, 2022, 42(2): 404-411. |
[9] | 李恒鑫, 常侃, 谭宇飞, 凌铭阳, 覃团发. 应用通道间相关性及增强信息蒸馏的彩色图像去马赛克网络[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 245-251. |
[10] | 卞凌志, 王直杰. 基于增强多维多粒度级联森林的信用评分模型[J]. 计算机应用, 2021, 41(9): 2539-2544. |
[11] | 黄梨, 卢龙. 基于长距离依赖编码与深度残差U-Net的缺血性卒中病灶分割[J]. 计算机应用, 2021, 41(6): 1820-1827. |
[12] | 牛康力, 谌雨章, 沈君凤, 曾张帆, 潘永才, 王绎冲. 基于深度学习的双通道夜视图像复原方法[J]. 计算机应用, 2021, 41(6): 1775-1784. |
[13] | 梁敏, 王昊榕, 张瑶, 李杰. 基于加速残差网络的图像超分辨率重建方法[J]. 计算机应用, 2021, 41(5): 1438-1444. |
[14] | 王海勇, 张开心, 管维正. 基于密集Inception的单图像超分辨率重建方法[J]. 《计算机应用》唯一官方网站, 2021, 41(12): 3666-3671. |
[15] | 曹建芳, 闫敏敏, 贾一鸣, 田晓东. 融合迁移学习的Inception-v3模型在古壁画朝代识别中的应用[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3219-3227. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||