FLAT: Chinese NER Using Flat-Lattice Transformer

ACL 2020  ·  Xiaonan Li, Hang Yan, Xipeng Qiu, Xuanjing Huang ·

Recently, the character-word lattice structure has been proved to be effective for Chinese named entity recognition (NER) by incorporating the word information. However, since the lattice structure is complex and dynamic, most existing lattice-based models are hard to fully utilize the parallel computation of GPUs and usually have a low inference-speed. In this paper, we propose FLAT: Flat-LAttice Transformer for Chinese NER, which converts the lattice structure into a flat structure consisting of spans. Each span corresponds to a character or latent word and its position in the original lattice. With the power of Transformer and well-designed position encoding, FLAT can fully leverage the lattice information and has an excellent parallelization ability. Experiments on four datasets show FLAT outperforms other lexicon-based models in performance and efficiency.

PDF Abstract ACL 2020 PDF ACL 2020 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Benchmark
Chinese Named Entity Recognition MSRA FLAT F1 94.12 # 13
Chinese Named Entity Recognition MSRA FLAT+BERT F1 96.09 # 5
Chinese Named Entity Recognition OntoNotes 4 FLAT F1 76.45 # 11
Chinese Named Entity Recognition OntoNotes 4 FLAT+BERT F1 81.82 # 6
Chinese Named Entity Recognition Resume NER FLAT F1 95.45 # 10
Chinese Named Entity Recognition Resume NER FLAT+BERT F1 95.86 # 6
Chinese Named Entity Recognition Weibo NER FLAT F1 60.32 # 14
Chinese Named Entity Recognition Weibo NER FLAT+BERT F1 68.55 # 8

Methods