Research map / Graph learning / Knowledge and temporal graphs
Bilinear Scoring Function Search for Knowledge Graph Learning
Yongqi Zhang, Quanming Yao, James Tin-Yau Kwok
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use evolutionary search instead of progressive greedy search | AutoBLM+ vs AutoBLM · FB15k | MRR: 0.853 -> 0.861 | Improved |
| use evolutionary search instead of progressive greedy search | AutoBLM+ vs AutoBLM · ogbl-wikikg2 | MRR: 0.532 -> 0.546 | Improved |
| apply a BLM searched on another dataset (FB15k237 to WN18RR) (ablation) | AutoBLM · WN18RR | MRR: 0.490 -> 0.462 | Hurt |
| apply a BLM searched on another dataset (WN18RR to FB15k237) (ablation) | AutoBLM · FB15k237 | MRR: 0.360 -> 0.345 | Hurt |
| add the performance predictor (ablation) | AutoBLM/AutoBLM+ · search on KG completion (validation top-8 MRR) | mean validation MRR vs time | Improved |
| use SRF features instead of one-hot features for predictor (ablation) | AutoBLM/AutoBLM+ · search on KG completion (validation top-8 MRR) | mean validation MRR vs time | Improved |
| add the filter that removes degenerate structures (ablation) | AutoBLM/AutoBLM+ · search on KG completion (validation top-8 MRR) | mean validation MRR vs time | Improved |
| increase initial nonzero count b0 from 1 to 4 without filter (ablation) | AutoBLM/AutoBLM+ · search on KG completion (validation top-8 MRR) | mean validation MRR vs time | Improved |
| use progressive/evolutionary search instead of Random, Bayes or Reinforce (ablation) | AutoBLM/AutoBLM+ · search on KG completion (validation top-8 MRR) | mean validation MRR vs time | Improved |
| increase K from 3 to 5 (ablation) | AutoBLM+ · search on KG completion (validation top-8 MRR) | best validation MRR | No clear effect |
| increase K from 3 to 5 (ablation) | AutoBLM+ · WN18RR | predictor running time (min): 1 -> 276 | Hurt |
| use parameter sharing (one-shot) to estimate structure quality (ablation) | AutoBLM+ · WN18RR | Spearman correlation with stand-alone MRR: -0.2686 | Hurt |
Closest papers by meaning
- Toward a Unified Statistical Theory of Unsupervised Pretraining and Supervised Neural Knowledge Graph Learning (NeurIPS 2026 · similarity 0.86)
- KGCRR: An Effective Metric-Driven Knowledge Graph Completion Framework by Designing a Novel Upper Bound Function with Adaptive Approximation to Reciprocal Rank (AAAI 2025 · similarity 0.86)
- Towards Global-Topology Relation Graph for Inductive Knowledge Graph Completion (AAAI 2025 · similarity 0.86)
- KGDM: A Diffusion Model to Capture Multiple Relation Semantics for Knowledge Graph Embedding (AAAI 2024 · similarity 0.85)
- Knowledge Graph Completion with Relation-Aware Anchor Enhancement (AAAI 2025 · similarity 0.85)
- Anchoring Path for Inductive Relation Prediction in Knowledge Graphs (AAAI 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
