A Neural Transition-based Model for Nested Mention Recognition
Abstract
It is common that entity mentions can contain other mentions recursively. This paper introduces a scalable transition-based method to model the nested structure of mentions. We first map a sentence with nested mentions to a designated forest where each mention corresponds to a constituent of the forest. Our shift-reduce based system then learns to construct the forest structure in a bottom-up manner through an action sequence whose maximal length is guaranteed to be three times of the sentence length. Based on Stack-LSTM which is employed to efficiently and effectively represent the states of the system in a continuous space, our system is further incorporated with a character-based component to capture letter-level patterns. Our model achieves the state-of-the-art results on ACE datasets, showing its effectiveness in detecting nested mentions.
Cite
@article{arxiv.1810.01808,
title = {A Neural Transition-based Model for Nested Mention Recognition},
author = {Bailin Wang and Wei Lu and Yu Wang and Hongxia Jin},
journal= {arXiv preprint arXiv:1810.01808},
year = {2018}
}
Comments
EMNLP 2018