LLM-based Atomic Propositions help weak extractors: Evaluation of a Propositioner for triplet extraction
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914442678108160 |
|---|---|
| author | Pommeret, Luc Gerald, Thomas Paroubek, Patrick Ghannay, Sahar Servan, Christophe Rosset, Sophie |
| author_facet | Pommeret, Luc Gerald, Thomas Paroubek, Patrick Ghannay, Sahar Servan, Christophe Rosset, Sophie |
| contents | Knowledge Graph construction from natural language requires extracting structured triplets from complex, information-dense sentences. In this paper, we investigate if the decomposition of text into atomic propositions (minimal, semantically autonomous units of information) can improve the triplet extraction. We introduce MPropositionneur-V2, a small multilingual model covering six European languages trained by knowledge distillation from Qwen3-32B into a Qwen3-0.6B architecture, and we evaluate its integration into two extraction paradigms: entity-centric (GLiREL) and generative (Qwen3). Experiments on SMiLER, FewRel, DocRED and CaRB show that atomic propositions benefit weaker extractors (GLiREL, CoreNLP, 0.6B models), improving relation recall and, in the multilingual setting, overall accuracy. For stronger LLMs, a fallback combination strategy recovers entity recall losses while preserving the gains in relation extraction. These results show that atomic propositions are an interpretable intermediate data structure that complements extractors without replacing them. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_02866 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LLM-based Atomic Propositions help weak extractors: Evaluation of a Propositioner for triplet extraction Pommeret, Luc Gerald, Thomas Paroubek, Patrick Ghannay, Sahar Servan, Christophe Rosset, Sophie Computation and Language Knowledge Graph construction from natural language requires extracting structured triplets from complex, information-dense sentences. In this paper, we investigate if the decomposition of text into atomic propositions (minimal, semantically autonomous units of information) can improve the triplet extraction. We introduce MPropositionneur-V2, a small multilingual model covering six European languages trained by knowledge distillation from Qwen3-32B into a Qwen3-0.6B architecture, and we evaluate its integration into two extraction paradigms: entity-centric (GLiREL) and generative (Qwen3). Experiments on SMiLER, FewRel, DocRED and CaRB show that atomic propositions benefit weaker extractors (GLiREL, CoreNLP, 0.6B models), improving relation recall and, in the multilingual setting, overall accuracy. For stronger LLMs, a fallback combination strategy recovers entity recall losses while preserving the gains in relation extraction. These results show that atomic propositions are an interpretable intermediate data structure that complements extractors without replacing them. |
| title | LLM-based Atomic Propositions help weak extractors: Evaluation of a Propositioner for triplet extraction |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2604.02866 |