Accurate Link Prediction for Edge-Incomplete Graphs via PU Learning (2405.11911v2)
Abstract: Given an edge-incomplete graph, how can we accurately find the missing links? The link prediction in edge-incomplete graphs aims to discover the missing relations between entities when their relationships are represented as a graph. Edge-incomplete graphs are prevalent in real-world due to practical limitations, such as not checking all users when adding friends in a social network. Addressing the problem is crucial for various tasks, including recommending friends in social networks and finding references in citation networks. However, previous approaches rely heavily on the given edge-incomplete (observed) graph, making it challenging to consider the missing (unobserved) links during training. In this paper, we propose PULL (PU-Learning-based Link predictor), an accurate link prediction method based on the positive-unlabeled (PU) learning. PULL treats the observed edges in the training graph as positive examples, and the unconnected node pairs as unlabeled ones. PULL effectively prevents the link predictor from overfitting to the observed graph by proposing latent variables for every edge, and leveraging the expected graph structure with respect to the variables. Extensive experiments on five real-world datasets show that PULL consistently outperforms the baselines for predicting links in edge-incomplete graphs.
- An enhanced recommender system based on heterogeneous graph link prediction. Engineering Applications of Artificial Intelligence 124 (2023), 106553.
- Seong-Jin Ahn and Myoung Ho Kim. 2021. Variational Graph Normalized AutoEncoders. In CIKM. ACM, 2827–2831.
- Lars Backstrom and Jure Leskovec. 2011. Supervised random walks: predicting and recommending links in social networks. In WSDM. ACM, 635–644.
- Albert-László Barabási and Réka Albert. 1999. Emergence of Scaling in Random Networks. Science (1999).
- Applications of link prediction in social networks: A review. J. Netw. Comput. Appl. 166 (2020), 102716.
- Maximum likelihood from incomplete data via the EM algorithm. Journal of the royal statistical society: series B (methodological) (1977).
- Analysis of Learning from Positive and Unlabeled Data. In NIPS. 703–711.
- Matthias Fey and Jan E. Lenssen. 2019. Fast Graph Representation Learning with PyTorch Geometric. In ICLR Workshop on Representation Learning on Graphs and Manifolds.
- Positive-Unlabeled Learning for Network Link Prediction. Mathematics 10, 18 (2022), 3345.
- Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable Feature Learning for Networks. In KDD. ACM, 855–864.
- Inductive Representation Learning on Large Graphs. In NIPS. 1024–1034.
- Yu Hao. 2021. Learning node embedding from graph structure and node attributes. Ph.D. Dissertation. UNSW Sydney.
- Thomas N. Kipf and Max Welling. 2016a. Semi-Supervised Classification with Graph Convolutional Networks. CoRR (2016).
- Thomas N. Kipf and Max Welling. 2016b. Variational Graph Auto-Encoders. CoRR abs/1611.07308 (2016).
- Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In ICLR (Poster). OpenReview.net.
- Positive-Unlabeled Learning with Non-Negative Risk Estimator. In NIPS. 1675–1685.
- A similarity-inclusive link prediction based recommender system approach. Elektronika IR Elektrotechnika 25, 6 (2019).
- Classifying networked text data with positive and unlabeled examples. Pattern Recognit. Lett. 77 (2016), 1–7.
- David Liben-Nowell and Jon M. Kleinberg. 2007. The link-prediction problem for social networks. J. Assoc. Inf. Sci. Technol. 58, 7 (2007), 1019–1031.
- Link prediction in paper citation network to construct paper correlation graph. EURASIP J. Wirel. Commun. Netw. 2019 (2019), 233.
- Pre-training graph neural networks for link prediction in biomedical networks. Bioinform. 38, 8 (2022), 2254–2262.
- Shuangxun Ma and Ruisheng Zhang. 2017. PU-LP: A novel approach for positive and unlabeled learning by label propagation. In ICME Workshops. IEEE Computer Society, 537–542.
- Kevin P Murphy. 2012. Machine learning: a probabilistic perspective. MIT press.
- Adversarially Regularized Graph Autoencoder for Graph Embedding. In IJCAI. ijcai.org, 2609–2615.
- Simon Parsons. 2011. Probabilistic Graphical Models: Principles and Techniques by Daphne Koller and Nir Friedman, MIT Press, 1231 pp., $95.00, ISBN 0-262-01319-3. Knowl. Eng. Rev. 26, 2 (2011), 237–238.
- DeepWalk: online learning of social representations. In KDD. ACM, 701–710.
- Link prediction in citation networks. J. Assoc. Inf. Sci. Technol. 63, 1 (2012), 78–85.
- Link prediction potentials for biological networks. Int. J. Data Min. Bioinform. 20, 2 (2018), 161–184.
- Graph Attention Networks. CoRR (2017).
- Structural Deep Network Embedding. In KDD. ACM, 1225–1234.
- Structural deep network embedding. In Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining. 1225–1234.
- Link prediction in social networks: the state-of-the-art. Sci. China Inf. Sci. 58, 1 (2015), 1–38.
- Long-short Distance Aggregation Networks for Positive Unlabeled Graph Learning. In CIKM. ACM, 2157–2160.
- Yiwei Xu. 2017. An Empirical Study of Locally Updated Large-scale Information Network Embedding (LINE). Ph.D. Dissertation. University of California, Los Angeles, USA.
- Accurate Graph-Based PU Learning without Class Prior. In ICDM. IEEE, 827–836.
- Positive and Unlabeled Learning with Label Disambiguation. In IJCAI. ijcai.org, 4250–4256.
- Han Zhang and Luyi Bai. 2023. Few-shot link prediction for temporal knowledge graphs based on time-aware translation and attention mechanism. Neural Networks 161 (2023), 371–381.
- Muhan Zhang and Yixin Chen. 2018. Link Prediction Based on Graph Neural Networks. In NeurIPS. 5171–5181.