Identifying Disease Outbreak Messages on Social Media Using Graph Neural Network Variants with TF-IDF and Word Embedding Features

Irwan Budiman; Mohammad Reza Faisal; Muliadi; Muhammad Rafi; Husnil Kamil; Luther Alexander Latumakulita; Luu Duc Ngo

doi:10.29207/resti.v10i5.7837

Authors

Irwan Budiman Lambung Mangkurat University
Mohammad Reza Faisal Lambung Mangkurat University
Muliadi Lambung Mangkurat University
Muhammad Rafi Lambung Mangkurat University
Husnil Kamil Andalas University
Luther Alexander Latumakulita Sam Ratulangi University
Luu Duc Ngo Bac Lieu University

DOI:

https://doi.org/10.29207/resti.v10i5.7837

Keywords:

deep learning, disease outbreak surveillance, feature extraction, social media, text classification

Abstract

The swift dissemination of information regarding disease outbreaks via social media presents significant challenges in the automatic identification of pertinent outbreak-related messages, particularly within multilingual textual data. The objective of this research is to devise and assess classification models grounded in Graph Neural Networks for the detection of disease outbreak messages within Indonesian and English datasets. The methodology employed encompasses text normalisation, feature extraction utilizing TF-IDF and word embedding techniques, including Word2Vec, fastText, and GloVe, followed by classification employing Graph Convolutional Networks (GCN) and Graph Attention Networks (GAT). The findings of this study indicate that the optimal performance on the Indonesian dataset was attained by GCN utilizing TF-IDF features amounting to 5000, resulting in an accuracy of 93%, whereas the superior GAT model attained 92% accuracy utilizing 4000 TF-IDF features. For the English dataset, the most effective GCN model reached an accuracy of 86.77% employing fastText, while the leading GAT model achieved 86.79% with the same fastText feature set. These results surpass those of previous investigations that employed CNN, LSTM, and BERT-based models for analogous tasks of disease outbreak message classification. This study concludes that GCN and GAT represent efficacious deep learning methodologies for the classification of multilingual disease outbreak messages. GCN demonstrates superior performance on Indonesian text utilizing TF-IDF-based features, while GAT exhibits marginally better performance on English text employing fastText features. In summary, graph-based models are capable of capturing relational information among textual features and delivering competitive classification performance when juxtaposed with earlier deep learning methodologies.

Downloads

Download data is not yet available.

References

[1] L. Tan, J. Guo, S. Mohanarajah, and K. Zhou, “Can we detect trends in natural disaster management with artificial intelligence? A review of modeling practices,” Natural Hazards, vol. 107, no. 3, pp. 2389–2417, 2021, doi: 10.1007/s11069-020-04429-3.

[2] S. Muppidi, P. S. Rao, and M. Rama Krishna Murthy, “Identification of natural disaster affected area using twitter,” in Advances in Decision Sciences, Image Processing, Security and Computer Vision, Springer, 2020, pp. 792–801, doi:10.1007/978-3-030-24322-7_92.

[3] R. Amalia et al., “Identification of Social Media Posts Containing Self-reported COVID-19 Symptoms using Triple Word Embeddings and Long Short-Term Memory,” Telematika, vol. 17, no. 1, pp. 27–38, 2024, doi: 10.35671/telematika.v17i1.2774.

[4] R. A. Rizian, I. Budiman, M. R. Faisal, D. Kartini, F. Indriani, and U. A. Ahmad, “Enhancing Classification of Self-Reported Monkeypox Symptoms on Social Media Using Term Frequency-Inverse Document Frequency Features and Graph Attention Networks,” Jurnal Teknik Informatika (Jutif), vol. 6, no. 6, pp. 5865–5881, 2025, doi: 10.52436/1.jutif.2025.6.6.5482.

[5] Y. Zhang, M. Ibaraki, and F. W. Schwartz, “Disease surveillance using online news: Dengue and zika in tropical countries,” J. Biomed. Inform., vol. 102, p. 103374, 2020, doi: 10.1016/j.jbi.2020.103374.

[6] R. Meckawy, D. Stuckler, A. Mehta, T. Al-Ahdal, and B. N. Doebbeling, “Effectiveness of early warning systems in the detection of infectious diseases outbreaks: a systematic review,” BMC Public Health, vol. 22, no. 1, p. 2216, 2022, doi:10.1186/s12889-022-14625-4.

[7] A. E. Wilson, C. U. Lehmann, S. N. Saleh, J. Hanna, and R. J. Medford, “Social media: a new tool for outbreak surveillance,” Antimicrobial Stewardship & Healthcare Epidemiology, vol. 1, no. 1, p. e50, 2021, doi:10.1017/ash.2021.225.

[8] Y. Liscano, L. A. Anillo Arrieta, J. F. Montenegro, D. Prieto-Alvarado, and J. Ordoñez, “Early warning of infectious disease outbreaks using social media and digital data: A scoping review,” Int. J. Environ. Res. Public Health, vol. 22, no. 7, p. 1104, 2025, doi:10.3390/ijerph22071104.

[9] R. Lamsal, M. R. Read, and S. Karunasekera, “CrisisTransformers: Pre-trained language models and sentence encoders for crisis-related social media texts,” Knowl. Based. Syst., vol. 296, p. 111916, 2024, doi:10.1016/j.knosys.2024.111916.

[10] A. Pak, A. Ziyaden, T. Saparov, I. Akhmetov, and A. Gelbukh, “Word Embeddings: A Comprehensive Survey,” Computacion y Sistemas, vol. 28, no. 4, pp. 2005 – 2029, 2024, doi: 10.13053/CyS-28-4-5225.

[11] Sutriawan, S. Rustad, G. F. Shidik, and Pujiono, “Performance Evaluation of Text Embedding Models for Ambiguity Classification in Indonesian News Corpus: A Comparative Study of TF-IDF, Word2Vec, FastText BERT, and GPT,” Ingenierie des Systemes d’Information, vol. 30, no. 6, pp. 1469 – 1482, 2025, doi:10.18280/isi.300606.

[12] M. Khairie, M. R. Faisal, R. Herteno, I. Budiman, F. Abadi, and M. I. Mazdadi, “The Effect of Channel Size on Performance of 1D CNN Architecture for Automatic Detection of Self-Reported COVID-19 Symptoms on Twitter,” in 2023 International Seminar on Intelligent Technology and Its Applications (ISITIA), 2023, pp. 621–625, doi:10.1109/ISITIA59021.2023.10220444.

[13] K. Yusupov, M. R. Islam, I. Muminov, M. Sahlabadi, and K. Yim, “Comparative Analysis of Machine Learning and Deep Learning Models for Email Spam Classification Using TF-IDF and Word Embedding Techniques,” Lecture Notes on Data Engineering and Communications Technologies, vol. 231, pp. 114 – 122, 2025, doi:10.1007/978-3-031-76452-3_11.

[14] C. M. Rohith, K. B. Sivachandra, N. Mohan, and S. S. Kumar, “Impact of Fine-Tuned Embeddings on Machine Learning-Based Causal Detection,” in 2025 5th International Conference on Intelligent Technologies, CONIT 2025, doi:10.1109/CONIT65521.2025.11167597.

[15] K. M. A. Kumar, S. G. Anusha, H. N. Sharath, M. Milind, and M. Charan, “Impact of Different Embeddings on Text Classification,” in 2024 International Conference on Recent Advances in Science and Engineering Technology (ICRASET), 2024, pp. 1–6, doi:10.1109/ICRASET63057.2024.10895636.

[16] A. M. Alayba and M. Altamimi, “Optimization of Arabic text classification using SVM integrated with word embedding models on a novel dataset,” International Journal of Advanced and Applied Sciences, vol. 12, no. 9, pp. 140 – 151, 2025, doi:10.21833/ijaas.2025.09.013.

[17] M. Danish and M. Amjad, Comparative Analysis of Word-Embedding Techniques Using LSTM Model. 2023. doi:10.1201/9781003371380-3.

[18] E. M. Dharma, F. L. Gaol, H. L. H. S. Warnars, and B. Soewito, “the Accuracy Comparison Among Word2Vec, Glove, and Fasttext Towards Convolution Neural Network (Cnn) Text Classification,” J. Theor. Appl. Inf. Technol., vol. 100, no. 2, pp. 349–359, 2022, doi:10.35870/jimik.v4i3.289.

[19] W. Zhao, L. Zhu, M. Wang, X. Zhang, and J. Zhang, “WTL-CNN: a news text classification method of convolutional neural network based on weighted word embedding,” Conn. Sci., vol. 34, no. 1, pp. 2291 – 2312, 2022, doi:10.1080/09540091.2022.2117274.

[20] M. A. B. Al-Tarawneh, O. Al-irr, K. S. Al-Maaitah, H. Kanj, and W. H. F. Aly, “Enhancing Fake News Detection with Word Embedding: A Machine Learning and Deep Learning Approach,” Computers, vol. 13, no. 9, 2024, doi:10.3390/computers13090239.

[21] S. Bodapati, H. Bandarupally, R. N. Shaw, and A. Ghosh, “Comparison and analysis of RNN-LSTMs and CNNs for social reviews classification,” in Advances in Applications of Data-Driven Computing, Springer, 2021, pp. 49–59.

[22] Z. Zhai, X. Zhang, F. Fang, and L. Yao, “Text classification of Chinese news based on multi-scale CNN and LSTM hybrid model,” Multimed. Tools Appl., pp. 1–14, 2023, doi:10.1007/s11042-023-14450-w.

[23] A. C. M. V. Srinivas, Ch. Satyanarayana, Ch. Divakar, and K. P. Sirisha, “Sentiment Analysis using Neural Network and LSTM,” IOP Conf. Ser. Mater. Sci. Eng., vol. 1074, no. 1, p. 012007, 2021, doi:10.1088/1757-899x/1074/1/012007.

[24] N. Rai, D. Kumar, N. Kaushik, C. Raj, and A. Ali, “Fake News Classification using transformer based enhanced LSTM and BERT,” International Journal of Cognitive Computing in Engineering, vol. 3, pp. 98–105, 2022, doi:10.1016/j.ijcce.2022.03.003.

[25] M. Khadhraoui, H. Bellaaj, M. Ben Ammar, H. Hamam, and M. Jmaiel, “Survey of BERT-base models for scientific text classification: COVID-19 case study,” Applied Sciences, vol. 12, no. 6, p. 2891, 2022, doi:10.3390/app12062891.

[26] I. Budiman et al., “Classification Performance Comparison of BERT and IndoBERT on SelfReport of COVID-19 Status on Social Media,” Journal of Computer Sciences Institute, vol. 30, pp. 61–67, 2024, doi:10.35784/jcsi.5564.

[27] N. R. Paul and R. C. Balabantaray, “Disaster related tweet classification method based on BERT and GAT,” International Journal of Information Technology, pp. 1–13, 2025, doi:10.1007/s41870-024-02389-6.

[28] Z. Chen, K. Huang, L. Wu, Z. Zhong, and Z. Jiao, “Relational Graph Convolutional Network for Text-Mining-Based Accident Causal Classification,” Applied Sciences, vol. 12, no. 5, 2022, doi:10.3390/app12052482.

[29] H. Ren et al., “Graph convolutional networks in language and vision: A survey,” Knowl. Based. Syst., vol. 251, 2022, doi:10.1016/j.knosys.2022.109250.

[30] S. M. Haider Rizvi, R. Imran, and A. Mahmood, “Text classification using graph convolutional networks: A comprehensive survey,” ACM Comput. Surv., vol. 57, no. 8, pp. 1–38, 2025, doi:10.1145/3714456.

[31] P. Pham, L. T. T. Nguyen, W. Pedrycz, and B. Vo, “Deep learning, graph-based text representation and classification: a survey, perspectives and challenges,” Artif. Intell. Rev., vol. 56, no. 6, pp. 4893–4927, 2023, doi:10.1007/s10462-022-10265-7.

[32] A. G. Vrahatis, K. Lazaros, and S. Kotsiantis, “Graph attention networks: a comprehensive review of methods and applications,” Future Internet, vol. 16, no. 9, p. 318, 2024, doi:10.3390/fi16090318.

[33] A. Z. Klein, A. Magge, K. O’Connor, J. I. Flores Amaro, D. Weissenbacher, and G. Gonzalez Hernandez, “Toward using Twitter for tracking COVID-19: a natural language processing pipeline and exploratory data set,” J. Med. Internet Res., vol. 23, no. 1, p. e25314, 2021, doi:10.2196/25314.

[34] A. Z. Klein, S. Kunatharaju, K. O’Connor, and G. Gonzalez-Hernandez, “Automatically identifying self-reports of COVID-19 diagnosis on Twitter: an annotated data set, deep neural network classifiers, and a large-scale cohort,” J. Med. Internet Res., vol. 25, p. e46484, 2023, doi:10.2196/46484.

[35] A. Tyagi, V. K. Jain, and V. Kumar, “Benchmark Text Preprocessing Techniques in Natural Language Processing,” in Proceedings - 2024 4th International Conference on Innovative Sustainable Computational Technologies, CISCT 2024, 2024, doi:10.1109/CISCT62494.2024.11134188.

[36] Y. Chai, Y. Liu, W. Li, B. Zhu, H. Liu, and Y. Jiang, “An interpretable wide and deep model for online disinformation detection,” Expert Syst. Appl., vol. 237, 2024, doi:10.1016/j.eswa.2023.121588.

[37] N. A. Semary, W. Ahmed, K. Amin, P. Pławiak, and M. Hammad, “Enhancing machine learning-based sentiment analysis through feature extraction techniques,” PLoS One, vol. 19, no. 2 February, 2024, doi:10.1371/journal.pone.0294968.

[38] Z. Shuai et al., “Comparison of different feature extraction methods for applicable automated ICD coding,” BMC Med. Inform. Decis. Mak., vol. 22, no. 1, 2022, doi:10.1186/s12911-022-01753-5.

[39] A. Pak, A. Ziyaden, T. Saparov, I. Akhmetov, and A. Gelbukh, “Word Embeddings: A Comprehensive Survey,” Computacion y Sistemas, vol. 28, no. 4, pp. 2005 – 2029, 2024, doi:10.13053/CyS-28-4-5225.

[40] M. N. Salim and B. S. Mustafa, “A Survey on Word Representation in Natural Language Processing,” in AIP Conference Proceedings, 2022, doi:10.1063/5.0121147.

[41] B. Samatha, N. Karyemsetty, D. S. Kumar, D. K. Rao, G. Mani, and T. Syamsundararao, “Analysis of a Multichannel Learning Mechanism for Speech Detection in Social Networks,” in Proceedings of the International Conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics, ICIITCEE 2023, 2023, pp. 1066 – 1071. doi:10.1109/IITCEE57236.2023.10091086.

[42] A. Balamurali and B. Ananthanarayanan, Develop a Neural Model to Score Bigram of Words Using Bag-of-Words Model for Sentiment Analysis, vol. 1–4. 2022, doi:10.4018/978-1-6684-6303-1.ch034.

[43] C. Sun et al., “Attention-based graph neural networks: a survey,” Artif. Intell. Rev., vol. 56, no. Suppl 2, pp. 2263–2310, 2023.

[44] G. Lachaud, P. Conde-Cespedes, and M. Trocan, “Mathematical Expressiveness of Graph Neural Networks,” Mathematics, vol. 10, no. 24, 2022, doi: 10.3390/math10244770.

[45] A. G. Vrahatis, K. Lazaros, and S. Kotsiantis, “Graph Attention Networks: A Comprehensive Review of Methods and Applications,” Future Internet, vol. 16, no. 9, 2024, doi: 10.3390/fi16090318.

[46] C. N. Tulu, “Experimental Comparison of Pre-Trained Word Embedding Vectors of Word2Vec, Glove, FastText for Word Level Semantic Text Similarity Measurement in Turkish,” Advances in Science and Technology Research Journal, vol. 16, no. 4, pp. 147 – 156, 2022, doi: 10.12913/22998624/152453.