Chem-FINESE: Validating Fine-Grained Few-shot Entity Extraction through Text Reconstruction

TL;DR

Chem-FINESE tackles the challenge of fine-grained, few-shot chemical entity extraction by coupling a seq2seq entity extractor with a seq2seq self-validation module that reconstructs the input sentence from extracted entities, guided by a decoder contrastive loss to discourage excessive copying. It introduces two new datasets, ChemNER+ and CHEMET, and demonstrates up to 8.26% and 6.84% absolute F1 improvements on these datasets, respectively, while also generalizing across the CrossNER domain without external knowledge or domain-adaptive pretraining. The approach explicitly targets statement-level faithfulness and long-tail type coverage, offering a practical, end-to-end training paradigm that leverages label semantics in generation. The work advances chemical information extraction by delivering robust, low-resource performance and providing publicly available benchmarks for future research.

Abstract

Fine-grained few-shot entity extraction in the chemical domain faces two unique challenges. First, compared with entity extraction tasks in the general domain, sentences from chemical papers usually contain more entities. Moreover, entity extraction models usually have difficulty extracting entities of long-tailed types. In this paper, we propose Chem-FINESE, a novel sequence-to-sequence (seq2seq) based few-shot entity extraction approach, to address these two challenges. Our Chem-FINESE has two components: a seq2seq entity extractor to extract named entities from the input sentence and a seq2seq self-validation module to reconstruct the original input sentence from extracted entities. Inspired by the fact that a good entity extraction system needs to extract entities faithfully, our new self-validation module leverages entity extraction results to reconstruct the original input sentence. Besides, we design a new contrastive loss to reduce excessive copying during the extraction process. Finally, we release ChemNER+, a new fine-grained chemical entity extraction dataset that is annotated by domain experts with the ChemNER schema. Experiments in few-shot settings with both ChemNER+ and CHEMET datasets show that our newly proposed framework has contributed up to 8.26% and 6.84% absolute F1-score gains respectively.

Figures (4)

• Figure 1: Comparison of sentence reconstruction results from ground truth and InBoXBART parmar-etal-2022-boxbart. We highlight Complete Correct, Missed Entity, and Partially Correct Prediction with different color.
• Figure 2: Type distributions for the training sets of ChemNER+ and CHEMET datasets. The Y-axis represents the number of mentions normalized by the mentions of the most frequent type. The X-axis represents the rank of types.
• Figure 3: Architecture overview. We use the example in Figure \ref{['fig:exp']} as a walking-through example.
• Figure 4: Average tokens in each mention for ChemNER+ and CHEMET datasets with few-shot settings.