Research map / Graph learning / Knowledge and temporal graphs
Logical Message Passing Networks with One-hop Inference on Atomic Formulas
Zihao Wang, Yangqiu Song, Ginny Wong, Simon See
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 LMPNN with pretrained ComplEx instead of ConE | complex query answering · FB15K | A_P MRR: 49.8 -> 50.6 | Improved |
| use LMPNN with pretrained ComplEx instead of Q2P | complex query answering · FB15K | A_N MRR: 16.4 -> 20.0 | Improved |
| use LMPNN with pretrained ComplEx instead of ConE | complex query answering · FB15K-237 | A_P MRR: 23.4 -> 24.1 | Improved |
| use LMPNN with pretrained ComplEx instead of Q2P | complex query answering · FB15K-237 | A_N MRR: 6.0 -> 7.8 | Improved |
| use LMPNN with pretrained ComplEx instead of ConE | complex query answering · NELL | A_P MRR: 27.2 -> 30.7 | Improved |
| use LMPNN with pretrained ComplEx instead of ConE | complex query answering · NELL | A_N MRR: 6.4 -> 8.0 | Improved |
| replace one-hop inference with a linear transformation of concatenated features (ablation) | LMPNN · FB15K-237 | A_P MRR: 24.1 -> 9.7 | Hurt |
| set ε to 0 instead of 0.1 (ablation) | LMPNN · FB15K-237 | A_P MRR: 24.1 -> 22.3 | Hurt |
| set ε to 0.5 instead of 0.1 (ablation) | LMPNN · FB15K-237 | A_P MRR: 24.1 -> 21.6 | Hurt |
| use one fewer message-passing layer than L (ablation) | LMPNN · FB15K-237 | A_P MRR: 24.1 -> 20.0 | Hurt |
| use one more message-passing layer than L (ablation) | LMPNN · FB15K-237 | A_P MRR: 24.1 -> 23.1 | Hurt |
| use two more message-passing layers than L (ablation) | LMPNN · FB15K-237 | A_P MRR: 24.1 -> 23.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ConE — ConE: Cone Embeddings for Multi-Hop Reasoning over Knowledge Graphs (NeurIPS 2021) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Inductive Logical Query Answering in Knowledge Graphs (NeurIPS 2022 · similarity 0.90)
- Efficient Embeddings of Logical Variables for Query Answering over Incomplete Knowledge Graphs (AAAI 2023 · similarity 0.90)
- Neural Methods for Logical Reasoning over Knowledge Graphs (ICLR 2022 · similarity 0.90)
- Rethinking Complex Queries on Knowledge Graphs with Neural Link Predictors (ICLR 2024 · similarity 0.90)
- Is Complex Query Answering Really Complex? (ICML 2025 · similarity 0.89)
- A Foundation Model for Zero-shot Logical Query Reasoning (NeurIPS 2024 · similarity 0.89)
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.
