Tree Structure LSTM for Chinese Named Entity Recognition

YUAN GAO, Hui Wang, Hongcheng Li

Keywords: LSTM, NER, dependency parsing, tree structure, Chinese

Abstract: In this paper, we remodel the bi-directional LSTM (Bi-LSTM) network for Chinese Named Entity Recognition (NER). We convert LSTM from chain-like structure into tree structure which is fixed to the dependency parsing tree of the sentence. The new structure model can fully leverage the syntax information of the sentence and has the capability of capturing long-range dependencies. In addition, we use dependency parsing label embedding to improve the performance of our approach. Experimental studies on four benchmarking Chinese NER datasets have verified the effectiveness of our approach.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes