{"slug": "schema-constrained-document-level-event-argument-extraction-with-lightweight-llm", "title": "Schema-Constrained Document-Level Event Argument Extraction with Lightweight LLM Fine-Tuning", "summary": "A new study from dessertlab shows that fine-tuned mid-sized open LLMs can outperform GPT baselines in schema-constrained document-level event argument extraction on the MAVEN-ARG benchmark. The approach combines role-set injection, LoRA fine-tuning, and deterministic decoding, with the best model (Phi-4, 14B) achieving 42.39% F1 at the event-coreference level. Code is publicly available at https://github.com/dessertlab/EAE/.", "body_md": "arXiv:2607.16808v1 Announce Type: new\nAbstract: Event Argument Extraction (EAE) converts documents into structured event records by identifying argument spans and assigning them schema-defined roles. Document-level EAE is challenging due to long-range dependencies between triggers and arguments, cross-sentence context, and strict role constraints, which often lead to boundary errors, uncertainty in roles, and inconsistencies with restricted schemas.\nIn this paper, we study whether mid-sized open LLMs can perform schema-constrained EAE reliably at the document level on MAVEN-ARG. Our approach combines (i) role-set injection in prompts for schema compliance, (ii) parameter-efficient supervised fine-tuning (LoRA) using the same JSON-only interface used at inference, and (iii) deterministic decoding with post-processing that validates JSON, filters invalid roles, de-duplicates arguments, and aligns spans to the document window. Under the official MAVEN-ARG evaluator, fine-tuned mid-sized open models outperform previously reported GPT baselines across mention, entity-coreference, and event-coreference evaluations; our best model (Phi-4, 14B) reaches 42.39\\% F1 at the event-coreference level. Code to reproduce experiments is publicly available at https://github.com/dessertlab/EAE/.", "url": "https://wpnews.pro/news/schema-constrained-document-level-event-argument-extraction-with-lightweight-llm", "canonical_source": "https://arxiv.org/abs/2607.16808", "published_at": "2026-07-21 04:00:00+00:00", "updated_at": "2026-07-21 04:23:11.966396+00:00", "lang": "en", "topics": ["natural-language-processing", "large-language-models", "ai-research"], "entities": ["dessertlab", "MAVEN-ARG", "Phi-4", "LoRA", "GPT"], "alternates": {"html": "https://wpnews.pro/news/schema-constrained-document-level-event-argument-extraction-with-lightweight-llm", "markdown": "https://wpnews.pro/news/schema-constrained-document-level-event-argument-extraction-with-lightweight-llm.md", "text": "https://wpnews.pro/news/schema-constrained-document-level-event-argument-extraction-with-lightweight-llm.txt", "jsonld": "https://wpnews.pro/news/schema-constrained-document-level-event-argument-extraction-with-lightweight-llm.jsonld"}}