Related Experiment Video
Updated: May 7, 2025
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
8.9K
Information extraction from green channel textual records on expressways using hybrid deep learning
Jiaona Chen1, Jing Zhang2, Weijun Tao2
1Xi'an Shiyou University School of Electronic Engineering, Xi'an, 710065, China. chenjn@xsyu.edu.cn.
Scientific Reports
|December 29, 2024
Summary
This study introduces a novel Named Entity Recognition (NER) model for China's expressway green channel, significantly improving failure case analysis. The RoBERTa-BiGRU-CRF model achieved high accuracy, enhancing agricultural logistics insights.
Area of Science:
- Agricultural Logistics
- Transportation Policy
- Natural Language Processing
Background:
- The expressway green channel is crucial for transporting fresh agricultural products in China.
- Analyzing failure cases within this system is vital for optimizing logistics.
- Existing methods for extracting information from textual failure records are limited.
Purpose of the Study:
- To develop a high-performance Named Entity Recognition (NER) model for extracting information from expressway green channel failure case records.
- To compare the effectiveness of different pre-trained natural language processing models for this specific task.
- To provide a systematic explanation of failure cases and insights for future research.
Main Methods:
- A hybrid approach combining BIO labeling, pre-trained models, deep learning, and Conditional Random Fields (CRF) was proposed.
- Eight key entities relevant to the expressway green channel were defined for NER.
- Three pre-trained models (BERT, ALBERT, RoBERTa) were utilized and compared for entity recognition and feature extraction.
Main Results:
- The RoBERTa-BiGRU-CRF model demonstrated superior performance with precision, recall, and F1-scores of 93.04%, 92.99%, and 92.99%, respectively.
- Text features extracted via pre-training significantly boosted the prediction accuracy of deep learning algorithms.
- The RoBERTa model proved highly effective for expressway green channel NER tasks.
Conclusions:
- Pre-trained language models, particularly RoBERTa, substantially enhance the accuracy of deep learning-based NER for agricultural logistics failure analysis.
- The developed NER model offers valuable knowledge extraction from textual data, providing systematic explanations and insights into expressway green channel failures.
- This research contributes to optimizing agricultural product transportation through improved data analysis and understanding of logistical challenges.
