Delving Deep into Regularity: A Simple but Effective Method for Chinese Named Entity Recognition
This addresses Chinese NER, a domain-specific task, by focusing on span regularity, offering a novel approach that improves performance over existing methods.
The paper tackles the problem of Chinese Named Entity Recognition by investigating the internal regularity of entity mentions, such as indicator words, and proposes a two-branch model that captures regularity for type prediction while locating boundaries. The method significantly outperforms previous state-of-the-art methods on three benchmark datasets and a practical medical dataset.
Recent years have witnessed the improving performance of Chinese Named Entity Recognition (NER) from proposing new frameworks or incorporating word lexicons. However, the inner composition of entity mentions in character-level Chinese NER has been rarely studied. Actually, most mentions of regular types have strong name regularity. For example, entities end with indicator words such as "company" or "bank" usually belong to organization. In this paper, we propose a simple but effective method for investigating the regularity of entity spans in Chinese NER, dubbed as Regularity-Inspired reCOgnition Network (RICON). Specifically, the proposed model consists of two branches: a regularity-aware module and a regularityagnostic module. The regularity-aware module captures the internal regularity of each span for better entity type prediction, while the regularity-agnostic module is employed to locate the boundary of entities and relieve the excessive attention to span regularity. An orthogonality space is further constructed to encourage two modules to extract different aspects of regularity features. To verify the effectiveness of our method, we conduct extensive experiments on three benchmark datasets and a practical medical dataset. The experimental results show that our RICON significantly outperforms previous state-of-the-art methods, including various lexicon-based methods.