Combined Self-Attention Mechanism for Chinese Named Entity Recognition in Military

Fei Liao,Liangli Ma,Jingjing Pei,Linshan Tan

doi:10.3390/fi11080180

Fei Liao, Liangli Ma + Show 2 more

Open Access

https://doi.org/10.3390/fi11080180

Copy DOI

Journal: Future Internet	Publication Date: Aug 18, 2019
Citations: 9	License type: CC BY 4.0

Affiliation: Naval University of Engineering

Abstract

Military named entity recognition (MNER) is one of the key technologies in military information extraction. Traditional methods for the MNER task rely on cumbersome feature engineering and specialized domain knowledge. In order to solve this problem, we propose a method employing a bidirectional long short-term memory (BiLSTM) neural network with a self-attention mechanism to identify the military entities automatically. We obtain distributed vector representations of the military corpus by unsupervised learning and the BiLSTM model combined with the self-attention mechanism is adopted to capture contextual information fully carried by the character vector sequence. The experimental results show that the self-attention mechanism can improve effectively the performance of MNER task. The F-score of the military documents and network military texts identification was 90.15% and 89.34%, respectively, which was better than other models.

Highlights

For the warring parties, how to extract a large amount of military information accurately in the shortest time will be the key link to win the initiative of war
To avoid cumbersome feature engineering and reduce the dependence on linguistic knowledge, our objective is to provide a neural network architecture that combines a bidirectional long short-term memory (BiLSTM) network with a self-attention mechanism to learn contextual features automatically
The experiment results show that our proposed model can improve the performance of Military named entity recognition (MNER) tasks

Summary

Introduction

How to extract a large amount of military information accurately in the shortest time will be the key link to win the initiative of war. Military named entity recognition (MNER) is a fundamental and important link of military information extraction used to detect military named entities (MNEs) from military text and classify them into predefined categories, such as troops, weapons, locations, missions, and organizations. It can extract valuable information from raw data to improve the efficiency of intelligence reconnaissance, command decision-making, organization, and implementation. Most research on MNER has employed statistical methods of machine learning, including conditional random fields (CRF) [2], the hidden Markov model (HMM) [3], and maximum entropy (ME) [4], and so forth These methods depend on handcrafted features and domain-specific knowledge resources excessively.

Objectives

Methods

Results

Conclusion