Noisy Positive-Unlabeled Learning with Self-Training for Speculative Knowledge Graph Reasoning
About
This paper studies speculative reasoning task on real-world knowledge graphs (KG) that contain both \textit{false negative issue} (i.e., potential true facts being excluded) and \textit{false positive issue} (i.e., unreliable or outdated facts being included). State-of-the-art methods fall short in the speculative reasoning ability, as they assume the correctness of a fact is solely determined by its presence in KG, making them vulnerable to false negative/positive issues. The new reasoning task is formulated as a noisy Positive-Unlabeled learning problem. We propose a variational framework, namely nPUGraph, that jointly estimates the correctness of both collected and uncollected facts (which we call \textit{label posterior}) and updates model parameters during training. The label posterior estimation facilitates speculative reasoning from two perspectives. First, it improves the robustness of a label posterior-aware graph encoder against false positive links. Second, it identifies missing facts to provide high-quality grounds of reasoning. They are unified in a simple yet effective self-training procedure. Empirically, extensive experiments on three benchmark KG and one Twitter dataset with various degrees of false negative/positive cases demonstrate the effectiveness of nPUGraph.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Knowledge Graph Reasoning | FB15k-237 (test) | HITS@3 (Avg)0.343 | 29 | |
| Link Prediction | Twitter ptb_rate = 0.3 (noisy and incomplete) | MRR3 | 13 | |
| Knowledge Graph Reasoning | FB15K ptb_rate=0.7 (test) | MRR36.5 | 12 | |
| Knowledge Graph Reasoning | FB15K-237 ptb_rate=0.7 (test) | MRR0.243 | 12 | |
| Knowledge Graph Reasoning | WN18 ptb_rate=0.7 (test) | MRR0.247 | 12 | |
| Knowledge Graph Reasoning | WN18 | MRR0.493 | 12 | |
| Knowledge Graph Reasoning | FB15K ptb_rate = 0.5 (test) | MRR0.417 | 12 | |
| Knowledge Graph Reasoning | FB15K-237 ptb_rate = 0.5 (test) | MRR0.258 | 12 | |
| Knowledge Graph Reasoning | WN18 ptb_rate = 0.5 (test) | MRR0.373 | 12 | |
| Knowledge Graph Reasoning | WN18 (test) | MRR0.63 | 12 |