Cross domain entity recognition improves type accuracy with confusion analysis
Mitigating Entity Type Confusion in Cross-Domain NER via Multidimensional Quantification and Reasoning Enhancement
Computation and LanguageArtificial Intelligence
Summary
Named Entity Recognition (NER) systems often get confused when identifying the exact type of an entity, especially when moving from one domain to another. The authors show that by carefully measuring how often and why these confusions happen, they can guide a language model to think through its classification in steps, reducing mistakes. Their approach significantly improves accuracy in recognizing different named entities across various topics. This means computers better understand what's being talked about, even when the subject changes.
What this means in practice
- •For natural language processing developers: Improve named entity type classification for domain-adaptive NER systems by reducing errors when transferring to new domains.
- •For enterprise chatbot teams: Enhance chatbot understanding by integrating improved entity recognition models that better distinguish similar entity types across topics.
Authors
Jingyu Wang, Shijie Wu, Fusheng Jin
Abstract
Cross-domain Named Entity Recognition (CD-NER) aims to transfer the rich knowledge in the source domain to the target domain. Recent studies adopting decomposition or generation paradigms have achieved significant performance improvements, demonstrating high accuracy in entity span detection. However, during entity type classification, models severely suffer from entity type confusion, the erroneous tendency that models classify entities of one type in the text as another similar but incorrect type. To address this issue, we first propose a Multidimensional Confusion Quantification Model (MCQM) that quantifies a model's confusion extent between entity types from three dimensions: source-target hierarchy analysis, semantic similarity analysis, and explicit data evaluation. Moreover, we propose the Progressive Bidirectional Reasoning Chain (PBRC). PBRC leverages the source-target hierarchy and confusion analysis from the MCQM to prompt the LLM to generate two-stage reasoning information. The two-stage reasoning information is utilized to augment the knowledge of the model, significantly mitigating entity type confusion and improving the model's generalization performance. Experimental results demonstrate that our method achieves new state-of-the-art results on all domains of the CrossNER dataset.