The week's pick
Reseach Article
Comparative Analysis of BERT and HGTCA-BERT Models for Disaster Tweet Classification
| International Journal of Computer Applications |
| Foundation of Computer Science (FCS), NY, USA |
| Volume 187 - Number 111 |
| Year of Publication: 2026 |
| Authors: Hitendra Kumar Prajapati, R.K. Sharma |
10.5120/ijca3b298097a6a5
|
Hitendra Kumar Prajapati, R.K. Sharma . Comparative Analysis of BERT and HGTCA-BERT Models for Disaster Tweet Classification. International Journal of Computer Applications. 187, 111 ( May 2026), 7-12. DOI=10.5120/ijca3b298097a6a5
Abstract
In recent years, social media applications like Twitter (X) have become important sources of real-time information in case of natural disasters. Nevertheless, the high velocity of unstructured and loud textual contents makes it difficult to precisely distinguish disaster-related information. The paper gives a comparative analysis of a baseline Bidirectional Encoder Representations from Transformers (BERT) model and a proposed hybrid model, HGTCA-BERT, on disaster tweet classification. The proposed HGTCA-BERT model integrates Hierarchical Graph (HG) structures to captures relationship between tweets, Temporal (T) features to model time-based information flow, and Cross-Attention (CA) mechanisms to enhance contextual understanding by combining textual, graph-based, and temporal representations. This multi-dimensional feature fusion enables the model to better understand the complex patterns present in disaster-related tweets. The experimental outcomes depict that HGTCA-BERT is better in performances, having an accuracy of (95%) much higher than BERT model (91%). This paper has demonstrated that HGTCA-BERT significantly outperforms the baseline BERT model by effectively leveraging structural and temporal dependencies alongside textual context.
References
- S. Imran, C. Castillo, F. Diaz, and S. Vieweg, “Processing Social Media Messages in Mass Emergency: A Survey,” ACM Computing Surveys, vol. 47, no. 4, pp. 1–38, 2015.
- S. Cresci, M. Tesconi, A. Cimino, and F. Dell’ Orletta, “Crisis Mapping during Disasters using Social Media,” Information Processing & Management, vol. 53, no. 3, pp. 559–572, 2017.
- D. Q. Nguyen, T. Vu, and A. T. Nguyen, “Improving Tweet Classification with Deep Neural Networks,” Proceedings of COLING, pp. 321–332, 2017.
- P. Liu, X. Qiu, and X. Huang, “Recurrent Neural Network for Text Classification with Multi-Task Learning,” IJCAI, pp. 2873–2879, 2018.
- J. Devlin, M. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” NAACL-HLT, pp. 4171–4186, 2019.
- Y. Liu et al., “RoBERTa: A Robustly Optimized BERT Pretraining Approach,” arXiv preprint arXiv:1907.11692, 2019.
- Z. Yang et al., “XLNet: Generalized Autoregressive Pretraining for Language Understanding,” NeurIPS, 2019.
- Z. Lan et al., “ALBERT: A Lite BERT for Self-supervised Learning of Language Representations,” ICLR, 2020.
- T. N. Kipf and M. Welling, “Semi-Supervised Classification with Graph Convolutional Networks,” ICLR, 2017.
- P.Veličković et al., “Graph Attention Networks,” ICLR, 2018.
- L. Yao, C. Mao, and Y. Luo, “Graph Convolutional Networks for Text Classification,” AAAI, pp. 7370–7377, 2019.
- J. Zhou et al., “Graph Neural Networks: A Review of Methods and Applications,” AI Open, vol. 1, pp. 57–81, 2020.
- S. M. Kazemi, R. Goel, K. Jain, et al., “Time2Vec: Learning a Vector Representation of Time,” arXiv:1907.05321, 2019.
- D. Xu, Y. Ruan, E. Korpeoglu, et al., “Inductive Representation Learning on Temporal Graphs,” ICLR, 2020.
- Z. Zhang et al., “Graph-BERT: Only Attention is Needed for Learning Graph Representations,” arXiv:2001.05140, 2021.
- Z. Wu et al., “A Comprehensive Survey on Graph Neural Networks,” IEEE Transactions on Neural Networks and Learning Systems, vol. 32, no. 1, pp. 4–24, 2021.
- J. Shang et al., “Temporal Graph Networks for Deep Learning on Dynamic Graphs,” ICML Workshop, 2022.
- S. N. Zisad, N. M. I. Chowdhury, and R. Hasan, “Comparative Analysis of Transformer Models in Disaster Tweet Classification for Public Safety,” arXiv preprint arXiv:2509.04650, 2025.
- A. Mehmood et al., “Disaster Tweet Classification using Transformer-based Models,” arXiv preprint arXiv:2405.00903, 2024.
- M. Naaz et al., “BERT-based Disaster Tweet Classification for Crisis Response,” EAI Endorsed Transactions on Scalable Information Systems, 2021.
- R. Garg et al., “BERT-based Models for Disaster Response and Summarization,” arXiv, 2023.
- A. Suleman et al., “Crisis Tweet Classification using Deep Learning Techniques,” arXiv, 2023.
- K. Dharrao, A. Singh, and P. Verma, “A Hybrid CNN Model with BERT Embeddings for Disaster Tweet Classification,” Journal of Information Processing Systems, vol. 20, no. 2, pp. 345–360, 2024.
- M. Baqi, R. Ahmed, and S. Khan, “Fusion of BERT and RoBERTa for Enhanced Text Classification,” International Journal of Advanced Computer Science, vol. 15, no. 3, pp. 210–220, 2024.
- S. Dinani, M. Raza, and T. Ali, “Comparative Study of BERT, LSTM, Bi-LSTM and CapsNet for Text Classification,” IEEE Access, vol. 11, pp. 56789–56802, 2023.
- R. Nath, P. Sharma, and V. Gupta, “BERT-GCN: A Hybrid Graph-based Model for Disaster Tweet Classification,” Expert Systems with Applications, vol. 235, 2025.
- Y. Li, X. Zhang, and H. Wang, “Graph Attention Networks for Social Media Text Classification,” IEEE Transactions on Neural Networks and Learning Systems, vol. 35, no. 4, pp. 1234–1245, 2024.
- J. Kim and S. Lee, “Context-Aware Disaster Tweet Classification using BERT and Graph Attention Networks,” Applied Soft Computing, vol. 140, 2025.
- D. Xu, Y. Ruan, E. Korpeoglu, et al., “Temporal and Dynamic Graph Learning for Social Media Analysis,” IEEE Transactions on Big Data, 2025.
- V. Stepanenko, “Disaster Tweets Dataset,” Kaggle, 2024.[Online]. Available: https://www.kaggle.com/datasets/vstepanenko/disaster-tweets
Index Terms
Keywords