ABSTRACT
In the actual world, many networks and graphs are fundamentally dynamic and heterogeneous. They contain many kinds of nodes and relations, and they are evolving with time. However, the majority of conventional link prediction techniques are currently limited to static or homogeneous networks and have the drawbacks of not fully exploiting the networks' time-domain evolutionary information as well as their extensive semantic and structural characteristics. In this paper, we propose a link prediction method (Att-ConvLSTM) that uses hierarchical attention to learn heterogeneous information and combines recurrent neural networks with temporal attention to capture evolutionary patterns. The proposed approach is found to be superior in AUC and Precision after comparison and analysis with various types of link predicting algorithms.
- Zhang M. Graph neural networks: link prediction. Graph Neural Networks: Foundations, Frontiers, and Applications, 2022: 195-223.Google Scholar
- Singer, U., Guy, I., Radinsky, K.: Node Embedding over Temporal Graphs. in:IJCAI. pp. 4605-4612 (2019)Google Scholar
- Pareja, A., Domeniconi, G., Chen, J., Ma, T., Suzumura, T., Kanezashi, H., Kaler,T., Schardl, T.B., Leiserson, C.E.: EvolveGCN: Evolving graph convolutional net-works for dynamic graphs. in: AAAI (2020)Google Scholar
- Sajadmanesh, S., Bazargani, S., Zhang, J., Rabiee, H.R.: Continuous-time rela-tionship prediction in dynamic heterogeneous information networks. ACM TKDD13(4), 44:1–44:31 (Jul 2019)Google Scholar
- Yin, Y., Ji, L., Zhang, J., Pei, Y.: Dhne: Network representation learning methodfor dynamic heterogeneous networks. IEEE Access 7, 134782–134792 (2019)Google ScholarCross Ref
- Dong Y, Liu Q, Du B, Weighted feature fusion of convolutional neural network and graph attention network for hyperspectral image classification. IEEE Transactions on Image Processing, 2022, 31: 1559-1572.Google ScholarDigital Library
- Shi X, Chen Z, Wang H, Convolutional LSTM network: A machine learning approach for precipitation nowcasting. Advances in Neural Information Processing Systems, 2015, 2015: 802-81Google Scholar
- Perozzi, B., Al-Rfou, R., Skiena, S.: DeepWalk: Online Learning of Social Repre-sentations. In: SIGKDD. pp. 701–710 (2014)Google Scholar
- Dong, Y., Chawla, N.V., Swami, A.: Metapath2Vec: Scalable RepresentationLearning for Heterogeneous Networks. In: SIGKDD. pp. 135–144 (2017)Google Scholar
- Sankar A, Wu Y, Gou L, Dynamic graph representation learning via self-attention networks. arXiv preprint arXiv:1812.09430, 2018.Google Scholar
- Xue H, Yang L, Jiang W, Modeling dynamic heterogeneous network for link prediction using hierarchical attention with temporal rnn//Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2020, Ghent, Belgium, September 14–18, 2020, Proceedings, Part I. Springer International Publishing, 2021: 282-298.Google Scholar
- Baytas I M, Xiao C, Wang F, Heterogeneous hyper-network embedding//2018 IEEE International Conference on Data Mining (ICDM). IEEE, 2018: 875-880.Google Scholar
Index Terms
- Dynamic Heterogeneous Link Prediction Based on Hierarchical Attention Model
Recommendations
Link Prediction in Heterogeneous Social Networks
CIKM '16: Proceedings of the 25th ACM International on Conference on Information and Knowledge ManagementA heterogeneous social network is characterized by multiple link types which makes the task of link prediction in such networks more involved. In the last few years collective link prediction methods have been proposed for the problem of link prediction ...
Modeling Dynamic Heterogeneous Network for Link Prediction Using Hierarchical Attention with Temporal RNN
Machine Learning and Knowledge Discovery in DatabasesAbstractNetwork embedding aims to learn low-dimensional representations of nodes while capturing structure information of networks. It has achieved great success on many tasks of network analysis such as link prediction and node classification. Most of ...
A supervised link prediction method for dynamic networks
Link prediction is an important sub-task in link mining area. This paper discusses link prediction in dynamic networks and proposes a new link prediction method which can learn from the long-term graph evolution of networks. The method first represents ...
Comments