Show Less, Instruct More: Enriching Prompts with Definitions and Guidelines for Zero-Shot NER
This addresses the problem of handling novel entity types in NER for NLP practitioners, though it is incremental as it builds on existing instruction-tuned LLMs.
The paper tackles zero-shot Named Entity Recognition (NER) for never-seen-before entity tags by proposing SLIMER, an approach that enriches prompts with definitions and guidelines, resulting in better performance, faster learning, and robustness, particularly for unseen entities, while performing comparably to state-of-the-art in out-of-domain settings.
Recently, several specialized instruction-tuned Large Language Models (LLMs) for Named Entity Recognition (NER) have emerged. Compared to traditional NER approaches, these models have demonstrated strong generalization capabilities. Existing LLMs primarily focus on addressing zero-shot NER on Out-of-Domain inputs, while fine-tuning on an extensive number of entity classes that often highly or completely overlap with test sets. In this work instead, we propose SLIMER, an approach designed to tackle never-seen-before entity tags by instructing the model on fewer examples, and by leveraging a prompt enriched with definition and guidelines. Experiments demonstrate that definition and guidelines yield better performance, faster and more robust learning, particularly when labelling unseen named entities. Furthermore, SLIMER performs comparably to state-of-the-art approaches in out-of-domain zero-shot NER, while being trained in a more fair, though certainly more challenging, setting.