Abstract:In recent years, the deep neural network has been introduced as an effective learning method in many natural language processing (NLP) applications. One of these applications is named entity recognition (NER), which is considered a vital role in the NLP systems (e.g., question/answering systems and translators). Since extracting entities traditionally requires massive computations to identify features manually (e.g., specific dictionaries), deep neural network methods have been introduced to overcome this challenge. In this work, we introduce a novel architecture that combines two different models of deep learning, namely convolutional neural network (CNN) and long short term memory (LSTM), to extract more efficient properties from an input sentence. The CNN extracts the local features of the individual words, and the LSTM network formulates the contextual information of the input sentence. In addition, thanks to an attention layer in our architecture, the performance has been improved. We implemented our experiments on two public datasets, CoNLL03 and ACE05. Evaluations demonstrate that employing the components of word-level CNN to capture local information of the input sentence and attention mechanism to focus on more relevant words leads to an enhancement in the performance of the NER system and establishes state-of-the-art results. Our architecture achieves F-score 92.00 and 86.34 on the two datasets CoNLL03 and ACE05, respectively. Comparing the previous works that utilize manually feature extraction computations or employ fewer components in their systems, the superiority of the proposed architecture in terms of accuracy is proven.

Method of Deep Web Entities Identification Based on BP Neural Network

Effective Approach to Deep Web Entries Identification

Character Recognition Based on Improved BP Neural Network

An automatic crisis information recognition model based on BP neural networks

Application of the Improved BP Neural Network in Ship and Ocean Engineering

Innovative Applications of BP Neural Networks in Financial Fraud Recognition

Research and Application on BP Neural Network Algorithm

Classification Algorithm of Network Public Opinion Text Information Based on BP Neural Network

Application Research of BP neural network in face recognition

Service Classification Using Adaptive Back-Propagation Neural Network And Semantic Similarity

Online biomedical named entities recognition by data and knowledge-driven model

Facial Expression Recognition System Based On Deep Residual Fusion Neural Network

Entity Relationship Extraction Based on Bi-LSTM and Attention Mechanism

BP Neural Network of Reservoir Identification Based on Moving-Weighted-average

Unlocking Deep Learning: A BP-Free Approach for Parallel Block-Wise Training of Neural Networks

Adaptive preprocessing of character recognition image based on neural network

Research on Recognition Method Based on Spectral Regression and Back Propagation Neural Network

Improving Feature Extraction Using a Hybrid of CNN and LSTM for Entity Identification

Language model based on deep learning network for biomedical named entity recognition

Face recognition fellow-up classification algorithm based on BP neural network

Brief Introduction of Back Propagation (BP) Neural Network Algorithm and Its Improvement