Link Prediction on N-ary Relational Facts: A Graph-based Approach
About
Link prediction on knowledge graphs (KGs) is a key research topic. Previous work mainly focused on binary relations, paying less attention to higher-arity relations although they are ubiquitous in real-world KGs. This paper considers link prediction upon n-ary relational facts and proposes a graph-based approach to this task. The key to our approach is to represent the n-ary structure of a fact as a small heterogeneous graph, and model this graph with edge-biased fully-connected attention. The fully-connected attention captures universal inter-vertex interactions, while with edge-aware attentive biases to particularly encode the graph structure and its heterogeneity. In this fashion, our approach fully models global and local dependencies in each n-ary fact, and hence can more effectively capture associations therein. Extensive evaluation verifies the effectiveness and superiority of our approach. It performs substantially and consistently better than current state-of-the-art across a variety of n-ary relational benchmarks. Our code is publicly available.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Hyper-Relational Link Prediction | JFFI100 V2 | H/T Score0.2957 | 22 | |
| Hyper-Relational Link Prediction | JFFI100 V1 | H/T Metric32.79 | 22 | |
| Hyper-Relational Link Prediction | WD20K66 V2 | H/T Score31.82 | 19 | |
| Hyper-Relational Link Prediction | WD20K33 V1 | H/T Score0.2062 | 19 | |
| Hyper-Relational Link Prediction | WD20K100 V2 | H/T Ratio20.36 | 19 | |
| Hyper-Relational Link Prediction | WD20K66 V1 | MRR (H/T)0.0176 | 19 | |
| Hyper-Relational Link Prediction | JFFI V1 | MRR (H/T)0.0487 | 18 | |
| Hyper-Relational Link Prediction | WD20K100 V1 | -- | 15 | |
| Entity Prediction | WikiPeople subject object | MRR50.3 | 14 | |
| Entity Prediction | JF17K subject/object | MRR0.617 | 14 |