TENER: Adapting Transformer Encoder for Named Entity Recognition

• DOI: 10.48550/arxiv.1911.04474
• Publication Date: 2019-01-01
• Authors: Hang Yan; Bocao Deng; Xiaonan Li; Xipeng Qiu
• Keywords: Named Entity Recognition

The Bidirectional long short-term memory networks (BiLSTM) have been widely used as an encoder in models solving the named entity recognition (NER) task. Recently, Transformer is broadly adopted various Natural Language Processing (NLP) tasks owing to its parallelism and advantageous performance. Nevertheless, performance of NER not good it other NLP tasks. In this paper, we propose TENER, a architecture adopting adapted Encoder model character-level features word-level features. By incorporating direction relative distance aware attention un-scaled attention, prove Transformer-like just effective for