论文题目: A Vectorized Relational Graph Convolutional Network for Multi-Relational Network Alignment
论文地址:https://www.ijcai.org/proceedings/2019/0574.pdf
研究动机
随着深度学习的发展,以图卷积网络(Graph Convolutional Network,GCN)[4]为代表的方法对图表示相关任务有很好的效果。然而,传统的基于频谱的GCN只能处理无向单关系网络,因为其要求归一化图拉普拉斯算子为实对称正半定矩阵,以便于进行图的傅立叶变换,这也表明邻接矩阵必须是对称的,并且二维的邻接矩阵也将边限制为相同类型,即单一关系。为了在图卷积中增加对多关系的支持,R-GCN[5]对每一关系学习一个映射矩阵,用于改变实体在累计邻居权重时考虑来自不同关系的影响,但R-GCN也没有显式地对关系进行表示。
a)显式的关系embedding学习。
b)实体角色区分:实体作为头实体或尾实体时采取不同的卷积操作,同时也体现图的有向性。
c)翻译模型的性质:学习到的表示具备形如TransE的h+r≈t的性质。
提出方法
1.VR-GCN框架
在VR-GCN的基础上,使用共享参数的设定分别训练两个输入的网络,同时增加网络对齐的目标,可以得到知识图谱对齐框架AVR-GCN。
实验分析
1.知识图谱对齐
评价指标使用的是MRR和Hits@k。两个指标越大,表明模型效果越好。
Baseline方面,使用了:将两个向量空间进行线性变换,从而达到对齐目的的MTransE[6];迭代式增加训练数据的ITransE[7];概率模型NTAM[8];使用受限的Margin ranking loss的BootEA[9]的非迭代模型AlignE;基于图卷积网络的实体对齐模型GCN-Align[10];以及AVR-GCN去除关系对齐目标的消融模型AVR-GCN(rl.exl.)。
可以看到AVR-GCN同样具有最好的效果。
这也符合我们的直观假设:训练数据越充分,模型效果越好。
数据集方面,使用的是链接预测任务广泛使用的公开数据集WN18和FB15k-237。
可以看到VR-GCN在提升网络表示效果上的提升还是很明显的。
IJCAI 2019涌现出很多实体对齐相关的文章,我们也将近年来基于表示学习的实体对齐方法做了整理,欢迎大家关注:https://github.com/THU-KEG/Entity_Alignment_Papers
[1] Bryan Perozzi, Rami Al-Rfou, and Steven Skiena. Deepwalk: online learning of social representations. In The 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’14, New York, NY, USA August 24 - 27, 2014, pages 701–710, 2014.
[2] Aditya Grover and Jure Leskovec. node2vec: Scalable feature learning for networks. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, pages 855–864, 2016.
[3] Antoine Bordes, Nicolas Usunier, Alberto Garc ́ıa-Dura ́n, Jason Weston, and Oksana Yakhnenko. Translating embeddings for modeling multi-relational data. In Advances in Neural Information Processing Systems 26: 27th Annual Conference on Neural Information Processing Systems 2013. Proceedings of a meeting held December 5-8, 2013, Lake Tahoe, Nevada, United States., pages 2787–2795, 2013.
[4] Thomas N. Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. CoRR, abs/1609.02907, 2016.
[5] Michael Sejr Schlichtkrull, Thomas N. Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, and Max Welling. Modeling relational data with graph convolutional networks. CoRR, abs/1703.06103, 2017.
[6] Muhao Chen, Yingtao Tian, Mohan Yang, and Carlo Zaniolo. Multilingual knowledge graph embeddings for cross-lingual knowledge alignment. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI 2017, Melbourne, Australia, August 19-25, 2017, pages 1511–1517, 2017.
[7] Hao Zhu, Ruobing Xie, Zhiyuan Liu, and Maosong Sun. Iterative entity alignment via joint knowledge embeddings. In Proceedings of the 26th International Joint Conference on Artificial Intelligence, IJCAI, Melbourne, Australia, pages 4258–4264, 2017.
[8] Shengnan Li, Xin Li, Rui Ye, Mingzhong Wang, Haiping Su, and Yingzi Ou. Non-translational alignment for multi-relational networks. In Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence, IJCAI 2018, July 13-19, 2018, Stockholm, Sweden., pages 4180–4186, 2018.
[9] Zequn Sun, Wei Hu, Qingheng Zhang, and Yuzhong Qu. Bootstrapping entity alignment with knowledge graph embedding. In Proceedings of the TwentySeventh International Joint Conference on Artificial Intelligence, IJCAI 2018, July 13-19, 2018, Stockholm, Sweden., pages 4396–4402, 2018.
[10] Zhichun Wang, Qingsong Lv, Xiaohan Lan, and Yu Zhang. Cross-lingual knowledge graph alignment via graph convolutional networks. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, Brussels, Belgium, October 31-November 4, 2018, pages 349–357, 2018.
[11] Bishan Yang, Wen-tau Yih, Xiaodong He, Jianfeng Gao, and Li Deng. Embedding entities and relations for learning and inference in knowledge bases. CoRR, abs/1412.6575, 2014.