Abstract:Fusing lexicon information into Chinese characters, which has normally a number of meanings, has been proven to be effective for Chinese Named Entity Recognition (NER). However, the existing approaches to incorporating a matched Chinese word into its composition characters only take the word as a whole (no subdivision or part), which failed to capture fine-grained correlation in word-character space and failed to make full use of lexicon information. Moreover, existing approaches use the fixed (static) weights between words and characters. This limits the performance of NER. Considering the fact that the same word-character pairs have different interactions in different contexts, the weights of matched word-character pairs should be dynamic rather than fixed. In this paper, we propose a Polymorphic Graph Attention Network (PGAT), aiming at capturing dynamic correlation between characters and matched words from multiple dimensions, to enhance the character representation. By obtaining matched words of characters from lexicon, we carefully map the word-character in four positions, which are B (begin), M (middle), E (end) and S (single word). The proposed semantic fusion unit based on Graph Attention Network (GAT) can dynamically modulate attention of matched words and characters in the four dimensions B, M, E, and S. Thus, it can explicitly capture fine-grained correlation between characters and matched words across each dimension. Experiments on four Chinese NER datasets show that PGAT outperforms the baseline models. It demonstrates the significance of the attention capture and fusion capabilities of the proposed polymorphic graph. Furthermore, PGAT is used in character representation layer, which makes it easier to be combined with pre-trained models like BERT and other sequence encoding models like CNN and Transformer.

A Multi-Channel Graph Attention Network for Chinese NER.

Polymorphic Graph Attention Network for Chinese NER

MGCN: A Novel Multi-Graph Collaborative Network for Chinese NER

A Local Information Perception Enhancement–Based Method for Chinese NER

MCSN: Multi-graph Collaborative Semantic Network for Chinese NER

MCL: Multi-Granularity Contrastive Learning Framework for Chinese NER

Dynamic Modeling Cross- and Self-Lattice Attention Network for Chinese NER

A Lexicon-Based Graph Neural Network for Chinese NER

CLGP: Multi-Feature Embedding Based Cross-Attention for Chinese NER

Porous Lattice-based Transformer Encoder for Chinese NER

Chinese named entity recognition based on Heterogeneous Graph and Dynamic Attention Network

Bi-Lattice LSTM Model with Self-Attention for Chinese NER

LSTM-CRF Neural Network with Gated Self Attention for Chinese NER

CGCL: A Novel Collaborative Graph Contrastive Learning Network for Chinese NER

Enhancing Chinese Character Representation with Lattice-Aligned Attention.

Pretraining Multi-modal Representations for Chinese NER Task with Cross-Modality Attention

Unified Lattice Graph Fusion for Chinese Named Entity Recognition

CSMA-CNER:Multi-modal Chinese NER Task with Cross- and Self-Modality Attention

Lattice LSTM for Chinese Sentence Representation

Bidirectional GRU with Multi-Head Attention for Chinese NER

Joint Self-Attention and Multi-Embeddings for Chinese Named Entity Recognition