journal article·2026-04-16Open access

Transformer-Based Emotion and Conflict Analysis of Disaster-Related Social Media: An Actor-Aware Decision Support Framework

Mesut Toğaçar, Serpil Aslan, Ayşe Meydanoğlu, Emirhan Denizyol, Abdurrezzak Ekidi, Tuncay Karateke, Yunus Emre Temiz, Beyzade Nadir Çetin, Ramazan Erten, Hatice Çakmak, Enes Saylan

Applied Sciences· Volume 16· Issue 8

Abstract

Social media platforms have become critical communication environments during disasters, where individuals express emotions, share information, and engage in public discourse. These platforms also reflect heterogeneous communication patterns shaped by different actor groups. However, existing studies predominantly focus on emotion classification and often overlook the combined role of actor identity and conflict dynamics. To address this gap, this study proposes an integrated AI-based analytical framework for actor-aware emotion and conflict analysis in post-disaster social media. An expert-annotated Turkish tweet dataset was constructed based on Ekman’s emotion model, including anger, fear, sadness, happiness, and surprise, along with an additional irrelevant/off-topic category and conflict-level labels. A Transformer-based model (BERTurk) was fine-tuned for multi-class emotion classification. Experimental results show that the proposed model achieves strong classification performance, with an accuracy of 0.931 and an F1-score of 0.912, outperforming conventional machine learning and deep learning baselines. Actor-based analysis reveals systematic differences in emotional and conflict patterns across groups. Scientists, journalists, and individual users exhibit higher levels of conflict and more pronounced negative emotional expressions, whereas institutionally oriented actors display comparatively balanced and supportive communication patterns. In addition, a web-based decision support system was developed to enable interactive visualization and actor-level exploration of emotional and conflict dynamics. Overall, the proposed framework provides a scalable, analytically robust approach to understanding social media discourse in disaster contexts and offers practical implications for AI-driven crisis communication and decision-support systems.
Source: crossref · Quality: baseline

Research context

OpenAlex topics: Sentiment Analysis and Opinion Mining · Public Relations and Crisis Communication · Mental Health via Writing

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

Affiliations

Department of Management Information Systems, Faculty of Economics and Administrative Sciences, Fırat University, 23119 Elazığ, Türkiye

Department of Software Engineering, Faculty of Engineering and Natural Sciences, Malatya Turgut Özal University, 44210 Malatya, Türkiye

Department of Basic Islamic Sciences, Arabic Language and Rhetoric Division, Faculty of Divinity, Fırat University, 23119 Elazığ, Türkiye

Department of Software Engineering, Institute of Graduate Education, Malatya Turgut Özal University, 44210 Malatya, Türkiye

Department of Philosophy and Religious Sciences, Religious Education Division, Faculty of Divinity, Fırat University, 23119 Elazığ, Türkiye

Department of Philosophy and Religious Studies, Religious Psychology Division, Faculty of Divinity, İnönü University, 44280 Malatya, Türkiye

Department of Sociology, Socio-Metrics Division, Faculty of Humanities and Social Sciences, Fırat University, 23119 Elazığ, Türkiye

Faculty of Human and Social Sciences, Modern Turkish Dialects and Literatures, Institute of Social Sciences, Fırat University, 23119 Elazığ, Türkiye

Department of Basic Islamic Sciences, Institute of Social Sciences, Fırat University, 23119 Elazığ, Türkiye

Funding

Scientific and Technological Research Council of Türkiye

References (43)

Show source-provided references
1. Accuracy of a pre-trained sentiment analysis classification model on tweets related to emergency response and early recovery assessment: The case of the 2019 Albanian earthquake10.1007/s11069-022-05307-w ↗
2. Contreras, D., Wilkinson, S., Fallou, L., Landès, M., Tomljenovich, I., Bossu, R., and James, P. (2021). Assessing emergency response and early recovery using sentiment analysis: The case of Zagreb, Croatia. Proceedings of the 1st Croatian Conference on Earthquake Engineering, Zagreb, Croatia, 22–24 March 2021, University of Zagreb.10.5592/CO/1CroCEE.2021.123 ↗
3. Real-time earthquake intensity estimation using streaming data analysis of social and physical sensors10.1007/s00024-016-1417-6 ↗
4. Effects of the 6 February 2023, Kahramanmaraş earthquake on structures in Kahramanmaraş city10.1007/s11069-023-06314-1 ↗
5. A novel attention-based deep learning model for improving sentiment classification after the case of the 2023 Kahramanmaraş/Turkey earthquake on Twitter10.7717/peerj-cs.2881 ↗
6. A deep learning-based sentiment analysis approach (MF-CNN-BiLSTM) and topic modeling of tweets related to the Ukraine–Russia conflict10.1016/j.asoc.2023.110404 ↗
7. Exploring aspect-based sentiment analysis: An in-depth review of current methods and prospects for advancement10.1007/s10115-024-02104-8 ↗
8. Deep learning for aspect-based sentiment analysis: A review10.7717/peerj-cs.1044 ↗
9. A novel sentiment analysis for response to natural disaster on Twitter data10.1142/S0218001423570070 ↗
10. Topic modelling and opinion analysis on climate change Twitter data using LDA and BERT model10.1016/j.procs.2023.01.071 ↗
11. Alqithami, S. (2025). Integrating Sentiment Analysis and Reinforcement Learning for Equitable Disaster Response: A Novel Approach. Sustainability, 17.10.3390/su17031072 ↗
12. Liu, B., and Zhang, L. (2012). A survey of opinion mining and sentiment analysis. Mining Text Data, Springer.10.1007/978-1-4614-3223-4_13 ↗
13. Plaza-del-Arco, F.M., Curry, A.A.C., Curry, A.C., and Hovy, D. (2024). Emotion analysis in NLP: Trends, gaps and roadmap for future directions. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), Torino, Italy, 20–25 May 2024, ELRA.10.63317/3kradf7tw6gm ↗
14. BMT-Net: Broad multitask transformer network for sentiment analysis10.1109/TCYB.2021.3050508 ↗
15. Review of recent emotion-annotated text corpora and resources10.1007/s10579-025-09828-1 ↗
16. Saraswathi, K., and Renukadevi, N.T. (2025). Multimodal Sentiment Analysis using Deep Learning and Transformers on Textual and Emojis based Social Media Inputs. Proceedings of the 2025 8th International Conference on Computing Methodologies and Communication (ICCMC), Erode, India, 29–31 July 2025, IEEE.10.1109/ICCMC65190.2025.11140993 ↗
17. Emotion-enriched word embeddings for Turkish10.1016/j.eswa.2023.120011 ↗
18. Context-aware sentiment analysis with attention-enhanced features from bidirectional transformers10.1007/s13278-022-00910-y ↗
19. Improving BERT model for requirements classification by bidirectional LSTM-CNN deep model10.1016/j.compeleceng.2023.108699 ↗
20. An argument for basic emotions10.1080/02699939208411068 ↗
21. Scherer, K.R., and Ekman, P. (2014). Expression and the nature of emotion. Approaches to Emotion, Psychology Press.10.4324/9781315798806 ↗
22. Urchs, S., Thurner, V., Aßenmacher, M., Heumann, C., and Thiemichen, S. (2025). Fair play in the newsroom: Actor-based filtering gender discrimination in text corpora. Proceedings of the 5th Workshop on Evaluation and Comparison of NLP Systems, St. Julians, Malta, 8–12 December 2025, ACL.10.18653/v1/2025.eval4nlp-1.5 ↗
23. Sentiment analysis in Turkish at different granularity levels10.1017/S1351324916000309 ↗
24. An enhanced guided LDA model augmented with BERT-based semantic strength for aspect term extraction in sentiment analysis10.1016/j.knosys.2022.108668 ↗
25. Analyzing tourism reviews using an LDA topic-based sentiment analysis approach10.1016/j.mex.2022.101894 ↗
26. Optimizing aspect-based sentiment analysis using sentence embedding transformer, Bayesian search clustering, and sparse attention mechanism10.1016/j.joitmc.2024.100211 ↗
27. Bhuvaneswari, A., and Kumudha, M. (2024). Topic modeling based clustering of disaster tweets using BERTopic. Proceedings of the 2024 MIT Art, Design and Technology School of Computing International Conference (MITADTSoCiCon), Chennai, India, 25–27 April 2024, IEEE.
28. Köksal, Ö (2021). Enhancing Turkish sentiment analysis using pre-trained language models. Proceedings of the 2021 29th Signal Processing and Communications Applications Conference (SIU), Istanbul, Turkey, 15–18 June 2021, IEEE.10.1109/SIU53274.2021.9477908 ↗
29. RapidAPI (2025, January 10). RapidAPI: The Largest API Hub. Available online: https://rapidapi.com/.
30. Apify (2025, January 11). Apify: Web Scraping and Automation Platform. Available online: https://apify.com/.
31. Using the Gini Index to quantify urban green inequality: A systematic review and recommended reporting standards10.1016/j.landurbplan.2024.105231 ↗
32. Inequalities between multi-rater kappas10.1007/s11634-010-0073-4 ↗
33. The 1994 Rwandan genocide: Practising the Heidelberg conflict methodology
34. Introducing the Heidelberg approach to conflict re-search10.1057/s41304-016-0093-4 ↗
35. Risch, J., Stoll, A., Wilms, L., and Wiegand, M. (2021). Overview of the Ger-mEval 2021 shared task on the identification of toxic, engaging, and fact-claiming com-ments. Proceedings of the GermEval 2021 Shared Task on the Identification of Toxic, Engaging, and Fact-Claiming Comments, ACL.
36. Affect, credibility, and solidarity: Strategic narratives of NGOs’ relief and advocacy efforts for Gaza10.1515/omgc-2024-0004 ↗
37. The Use of Dysphemism in the Comments Column on the Instagram Account @stateofisrael
38. Discursive strategies of blaming: The lan-guage of judgment and political protest online10.1177/20563051221138753 ↗
39. Chouikhi, H., Chniter, H., and Jarray, F. (2021). Arabic sentiment analysis using BERT model. Proceedings of the International Conference on Computational Collective Intelligence, Rhodes, Greece, 29 September–1 October 2021, Springer.10.1007/978-3-030-88113-9_50 ↗
40. A novel TCNN–Bi-LSTM deep learning model for predicting sentiments of tweets about COVID-19 vaccines10.1002/cpe.7387 ↗
41. (2026, February 10). Conflict Language Analysis Platform (AI-Based Social Media Analytics System). Available online: https://catismadili.com/.
42. Impact of random oversampling and random undersampling on the performance of prediction models developed using observational health data10.1186/s40537-023-00857-7 ↗
43. Improving text classification with weighted word embeddings via a multi-channel TextCNN model10.1016/j.neucom.2019.07.052 ↗