sis 18: e8

Research Article

Sequence Classification of Tweets with Transfer Learning via BERT in the Field of Disaster Management

Download91 downloads
  • @ARTICLE{10.4108/eai.23-3-2021.169071,
        author={Sumera Naaz and Zain Ul Abedin and Danish Raza Rizvi},
        title={Sequence Classification of Tweets with Transfer Learning via BERT in the Field of Disaster Management},
        journal={EAI Endorsed Transactions on Scalable Information Systems: Online First},
        volume={},
        number={},
        publisher={EAI},
        journal_a={SIS},
        year={2021},
        month={3},
        keywords={BERT (Bidirectional Encoder Representation from Transformers), Tweet classification, Balanced Dataset, Imbalanced Dataset, Disaster Management, Natural Language Processing},
        doi={10.4108/eai.23-3-2021.169071}
    }
    
  • Sumera Naaz
    Zain Ul Abedin
    Danish Raza Rizvi
    Year: 2021
    Sequence Classification of Tweets with Transfer Learning via BERT in the Field of Disaster Management
    SIS
    EAI
    DOI: 10.4108/eai.23-3-2021.169071
Sumera Naaz1, Zain Ul Abedin1, Danish Raza Rizvi1,*
  • 1: Department of Computer Engineering, Jamia Millia Islamia, New Delhi,110025 India
*Contact email: Drizvi@jmi.ac.in

Abstract

Twitter is extensively used as an information-sharing platform during any kind of emergency like disasters etc. People tweet useful information about disaster-related events such as evacuations, volunteer need, help, warnings etc. This data is sometimes very useful for rescue teams, NGOs, military and various other government and private organisations who are tasked with responsibilities to save lives and provide volunteers. This data can also be used to analyze disaster behaviour. In this paper, we have collected labelled tweets from crisisLexT26 and crisisNLP and classified them into seven labels on the basis of information provided by them. The data was heavily skewed. So to improve the accuracy of classifiers, we have applied various techniques as a result of which we have created two datasets (Imbalanced and Balanced). We have compared the performance of various BERT-based models on these two datasets. For sequence classification, a balanced dataset performs better than an imbalanced dataset. We can improve accuracy of classifiers to great extent by adopting good data preprocessing and data splitting techniques.