Entity Candidate Network for Whole-Aware Named Entity Recognition

Wendong He, Yizhen Shao, Pingjian Zhang

Named Entity Recognition (NER) is a crucial upstream task in Natural Language Processing (NLP). Traditional tag scheme approaches offer a single recognition that does not meet the needs of many downstream tasks such as coreference resolution. Meanwhile, Tag scheme approaches ignore the continuity of entities. Inspired by one-stage object detection models in computer vision (CV), this paper proposes a new no-tag scheme, the Whole-Aware Detection, which makes NER an object detection task. Meanwhile, this paper presents a novel model, Entity Candidate Network (ECNet), and a specific convolution network, Adaptive Context Convolution Network (ACCN), to fuse multi-scale contexts and encode entity information at each position. ECNet identifies the full span of a named entity and its type at each position based on Entity Loss. Furthermore, ECNet is regulable between the highest precision and the highest recall, while the tag scheme approaches are not. Experimental results on the CoNLL 2003 English dataset and the WNUT 2017 dataset show that ECNet outperforms other previous state-of-the-art methods.

Knowledge Graph

arrow_drop_up

Comments

Sign up or login to leave a comment