Research map / Large language models / Retrieval and knowledge
Adapting Neural Link Predictors for Data-Efficient Complex Query Answering
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Adapt neural link prediction scores with a learned calibration layer | ComplEx-N3 with CQD · All datasets and query types | MRR: 34.4 to 35.1 | Helped |
| Adapt link prediction scores | ComplEx-N3 with CQD · FB15K test queries with negation | MRR: 38.6 -> 42.8 | Helped |
| Adapt link prediction scores | ComplEx-N3 with CQD · FB15K-237 test EPFO queries | MRR: 26.8 -> 25.7 | Hurt |
| Adapt link prediction scores | ComplEx-N3 with CQD · FB15K-237 test queries with negation | MRR: 10.2 -> 10.7 | Helped |
| Adapt link prediction scores | ComplEx-N3 with CQD · NELL995 test EPFO queries | MRR: 28.9 -> 32.3 | Helped |
| Adapt link prediction scores | ComplEx-N3 with CQD · NELL995 test queries with negation | MRR: 9.7 -> 13.3 | Helped |
| Use only 1% of complex-query training examples (ablation) | Ours · FB15K-237 test | MRR: 2.2 | Hurt |
| Train on 1% of 2i queries only (ablation) | Ours · Unseen complex query types | MRR: 2.9 | Hurt |
| Train on 1% of 2i queries only (ablation) | Ours · Queries containing negations | MRR: 2.7 | Hurt |
| Fine-tune all link predictor parameters instead of keeping the predictor frozen and adapting scores (ablation) | CQD and Ours · FB15K-237 test queries | MRR: 13.2 vs 9.3 | Hurt |
| Replace initial entity and relation representations with learned transformations (ablation) | CQD · FB15K-237 test, 2p query | MRR: 13.2 -> 6.4 | Hurt |
| Concatenate learned transformations to initial entity and relation representations (ablation) | CQD · FB15K-237 test, 2p query | MRR: 13.2 -> 10.9 | Hurt |
Closest papers by meaning
- Complex Query Answering with Neural Link Predictors (ICLR 2021 · similarity 0.95)
- Efficient Embeddings of Logical Variables for Query Answering over Incomplete Knowledge Graphs (AAAI 2023 · similarity 0.91)
- Rethinking Complex Queries on Knowledge Graphs with Neural Link Predictors (ICLR 2024 · similarity 0.89)
- Is Complex Query Answering Really Complex? (ICML 2025 · similarity 0.88)
- Listening to the Wise Few: Query–Key Alignment Unlocks Latent Correct Answers in Large Language Models (NeurIPS 2026 · similarity 0.85)
- A Foundation Model for Zero-shot Logical Query Reasoning (NeurIPS 2024 · similarity 0.85)
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.