Extraction of polymer biodegradation events from scientific papers: structured dataset design and contradiction-aware scientific event assessment

Scientific literature on polymer biodegradation contains experimentally important information regarding degradable polymers, microorganisms, environmental conditions, and degradation outcomes. However, biodegradation findings are often condition-sensitive and distributed across full-text articles, making reliable event extraction and evaluation difficult. In this work, we introduce PolyBD-EE, a document-level polymer biodegradation event extraction dataset with structured event annotations and text-grounded participant expansion. We further propose a contradiction-aware evaluation framework that evaluates whether predicted and gold events describe experimentally compatible biodegradation findings under different participant specificity levels. Using PolyBD-EE, we benchmark multiple modern generative large language models under zero-shot extraction settings. The best-performing setting, GPT-5.5-Think, achieves strict- and lenient-atomic F1 scores of 0.492 and 0.557, respectively, indicating that current models can capture underlying biodegradation findings but still struggle with fine-grained participant specificity described in the literature. Further analysis identifies key bottlenecks, including incorrect participant configurations, over-generalization during participant expansion, and weak separation of event-specific information. Manual validation of GPT-5.5-Think predictions confirms 96.9% and 96.6% of accepted alignments under strict and lenient evaluation, respectively, with false-rejection rates of 0.55% and 1.85%. These findings show that document-level biodegradation event extraction remains a challenging task with substantial room for improvement. PolyBD-EE and the proposed evaluation framework provide a task-specific basis for measuring future progress, diagnosing failure modes, and guiding the development of more reliable scientific event extraction systems for polymer biodegradation research and experimental materials science. Code and publicly redistributable data are available at: https://github.com/sl-633/polybd-ee-caef