RADAR: Reasoning as Discrimination with Aligned Representations for LLM-based Knowledge Graph Reasoning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914350439071744 |
|---|---|
| author | Xue, Bo Jin, Yuan Fu, Luoyi Ding, Jiaxin Wang, Xinbing |
| author_facet | Xue, Bo Jin, Yuan Fu, Luoyi Ding, Jiaxin Wang, Xinbing |
| contents | Knowledge graph reasoning (KGR) infers missing facts, with recent advances increasingly harnessing the semantic priors and reasoning abilities of Large Language Models (LLMs). However, prevailing generative paradigms are prone to memorizing surface-level co-occurrences rather than learning genuine relational semantics, limiting out-of-distribution generalization. To address this, we propose RADAR, which reformulates KGR from generative pattern matching to discriminative relational reasoning. We recast KGR as discriminative entity selection, where reinforcement learning enforces relative entity separability beyond token-likelihood imitation. Leveraging this separability, inference operates directly in representation space, ensuring consistency with the discriminative optimization and bypassing generation-induced hallucinations. Across four benchmarks, RADAR achieves 5-6% relative gains on link prediction and triple classification over strong LLM baselines, while increasing task-relevant mutual information in intermediate representations by 62.9%, indicating more robust and transferable relational reasoning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_21951 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | RADAR: Reasoning as Discrimination with Aligned Representations for LLM-based Knowledge Graph Reasoning Xue, Bo Jin, Yuan Fu, Luoyi Ding, Jiaxin Wang, Xinbing Computation and Language Knowledge graph reasoning (KGR) infers missing facts, with recent advances increasingly harnessing the semantic priors and reasoning abilities of Large Language Models (LLMs). However, prevailing generative paradigms are prone to memorizing surface-level co-occurrences rather than learning genuine relational semantics, limiting out-of-distribution generalization. To address this, we propose RADAR, which reformulates KGR from generative pattern matching to discriminative relational reasoning. We recast KGR as discriminative entity selection, where reinforcement learning enforces relative entity separability beyond token-likelihood imitation. Leveraging this separability, inference operates directly in representation space, ensuring consistency with the discriminative optimization and bypassing generation-induced hallucinations. Across four benchmarks, RADAR achieves 5-6% relative gains on link prediction and triple classification over strong LLM baselines, while increasing task-relevant mutual information in intermediate representations by 62.9%, indicating more robust and transferable relational reasoning. |
| title | RADAR: Reasoning as Discrimination with Aligned Representations for LLM-based Knowledge Graph Reasoning |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2602.21951 |