Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Learning to Walk with Dual Agents for Knowledge Graph Reasoning
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 |
|---|---|---|---|
| Use dual-agent cluster-level guidance and mutual reinforcement instead of entity-only walking | CURL versus MINERVA · FB15K-237, WN18RR, NELL-995 | Hits@1: 22.4 vs 19.2; 42.9 vs 41.3; 66.7 vs 58.8 | Mixed |
| Use dual-agent cluster-level guidance and mutual reinforcement instead of entity-only walking | CURL versus MINERVA · FB15K-237, WN18RR, NELL-995 | MRR: 30.6 vs 27.1; 46.0 vs 44.8; 73.8 vs 67.5 | Mixed |
| Increase cluster count from 50 to 75 (ablation) | CURL · Four NELL-995 relation tasks | Hits@10 | Helped |
| Increase cluster count beyond 75, up to 500 (ablation) | CURL · Four NELL-995 relation tasks | Hits@10 | Hurt |
| Increase beam size from 20 to 50 (ablation) | CURL · OrganizationHeadquarteredInCity, NELL-995 | Hits@1: 74.7 -> 75.9 | Helped |
| Increase beam size from 50 to 100 (ablation) | CURL · OrganizationHeadquarteredInCity, NELL-995 | Hits@1: 75.9 -> 76.3 | No clear effect |
Closest papers by meaning
- Walk Wisely on Graph: Knowledge Graph Reasoning with Dual Agents via Efficient Guidance-Exploration (AAAI 2025 · similarity 0.94)
- Knowledge-Graph Paths as Intermediate Supervision for Self-Evolving Search Agents (NeurIPS 2026 · similarity 0.87)
- Search-R2: Enhancing Search-Integrated Reasoning via Actor-Refiner Collaboration (ICML 2026 · similarity 0.87)
- $\texttt{G1}$: Teaching LLMs to Reason on Graphs with Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Plan-on-Graph: Self-Correcting Adaptive Planning of Large Language Model on Knowledge Graphs (NeurIPS 2024 · similarity 0.87)
- LoongRL: Reinforcement Learning for Advanced Reasoning over Long Contexts (ICLR 2026 · similarity 0.87)
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.