TRI-EMBEDDINGS: A NOVEL APPROACH FOR DETECTING ABUSIVE LANGUAGE ON SOCIAL MEDIA

Rezan Bakır

doi:10.17780/ksujes.1679361

TR EN

TRI-EMBEDDINGS: SOSYAL MEDYADA SALDIRGAN DİL TESPİTİ İÇİN YENİ BİR YAKLAŞIM

Öz

Küresel dijital iletişimin artışı, nefret söylemi ve uygunsuz dil içeren metinlerin yayılmasına yol açmış ve bu durum toplumsal refahı tehdit etmektedir. Yapay zekâ, siber güvenlik zorlukları yaratırken, aynı zamanda bu sorunların çözülmesinde de kritik bir rol oynamaktadır. Araştırmacılar, algoritmaların yanlış kullanımını azaltmak ve siber güvenliği sağlamak için çok disiplinli bir yaklaşım benimsemelidir. Bu çalışmada, Twitter verisi üzerinde yapay zekâ destekli metin analiziyle uygunsuz kelimeler içeren metni tespit etmek amacıyla Tri-Gömülü Temsiller (Tri-Embeddings) adı verilen yenilikçi bir yöntem sunulmaktadır. Bu yöntem, Word2Vec, FastText ve Universal Sentence Encoder (USE) gibi önceden eğitilmiş modelleri birleştirir. Semantik incelikleri yakalamak ve modelin uygunsuz kelimeler içeren metindeki ince farkları tespit etme yeteneğini artırmak için yapılan bu entegrasyon, doğruluk ve sağlamlık açısından önemli gelişmeler sağlamıştır. Ek olarak, DistilBERT gibi büyük bir dil modelinin önerilen yönteme dâhil edilmesinin etkisi de incelenmiştir. Bulgular, yüksek doğruluk, duyarlılık ve F1 skoru değerleriyle yöntemin etkinliğini göstermektedir. Bu yaklaşım, etik ihlalleri azaltarak daha güvenli ve kapsayıcı bir çevrimiçi ortam yaratmaya yardımcı olmaktadır.

Anahtar Kelimeler

TRI-EMBEDDINGS: A NOVEL APPROACH FOR DETECTING ABUSIVE LANGUAGE ON SOCIAL MEDIA

Abstract

The rise in global digital communication has led to an increase in hate speech and offensive language, posing significant threats to societal well-being. While AI presents cybersecurity challenges, it also plays a crucial role in addressing these issues. Researchers must develop AI with a multidisciplinary approach, mitigating algorithmic misuse and ensuring cybersecurity. This study introduces Tri-Embeddings, an innovative method for detecting abusive language using AI-powered text analysis, applied to Twitter data. The method combines pre-trained models such as Word2Vec, FastText, and Universal Sentence Encoder (USE). Additionally, this study explores the impact of integrating a large language model, DistilBERT, into the proposed unified embedding framework. The findings demonstrate the method’s effectiveness, with high precision, recall, and F1 scores, showing its potential to reduce the spread of offensive and hateful language. This approach helps mitigate ethical breaches and creates a safer, more inclusive online space.

Keywords

References

Al-Saqqa, S., & Awajan, A. (2019). The use of word2vec model in sentiment analysis: A survey. Proceedings of the 2019 International Conference on Artificial Intelligence, Robotics and Control, 39–43. https://doi.org/10.1145/3388218.3388229
Barbieri, F., Camacho-Collados, J., Neves, L., & Espinosa-Anke, L. (2020). TweetEval: Unified benchmark and comparative evaluation for tweet classification. ArXiv Preprint ArXiv:2010.12421. https://doi.org/10.18653/v1/2020.findings-emnlp.148
Bojanowski, P., Grave, E., Joulin, A., & Mikolov, T. (2017). Enriching word vectors with subword information. Transactions of the Association for Computational Linguistics, 5, 135–146. https://doi.org/10.1162/tacl_a_00051
Cécillon, N., Labatut, V., Dufour, R., & Linares, G. (2021). Graph embeddings for abusive language detection. SN Computer Science, 2, 1–15. https://doi.org/10.1007/s42979-020-00413-7
Cer, D., Yang, Y., Kong, S., Hua, N., Limtiaco, N., John, R. S., Constant, N., Guajardo-Cespedes, M., Yuan, S., & Tar, C. (2018). Universal sentence encoder. ArXiv Preprint ArXiv:1803.11175. https://doi.org/10.18280/ria.350404
Davidson, T., Bhattacharya, D., & Weber, I. (2019a). Racial bias in hate speech and abusive language detection datasets. ArXiv Preprint ArXiv:1905.12516. https://doi.org/10.18653/v1/w19-3504
Davidson, T., Warmsley, D., Macy, M., & Weber, I. (2017). Automated hate speech detection and the problem of offensive language. Proceedings of the International AAAI Conference on Web and Social Media, 11(1), 512–515. https://doi.org/10.1609/icwsm.v11i1.14955
Dorris, W., Hu, R., Vishwamitra, N., Luo, F., & Costello, M. (2020). Towards automatic detection and explanation of hate speech and offensive language. Proceedings of the Sixth International Workshop on Security and Privacy Analytics, 23–29. https://doi.org/10.1145/3375708.3380312

Farha, I. A., & Magdy, W. (2020). Multitask learning for Arabic offensive language and hate-speech detection. Proceedings of the 4th Workshop on Open-Source Arabic Corpora and Processing Tools, with a Shared Task on Offensive Language Detection, 86–90.
Fetahi, E., Susuri, A., Hamiti, M., Kastrati, Z., Canhasi, E., & Misini, A. (2025). Enhancing social media hate speech detection in low-resource languages using transformers and explainable AI. Social Network Analysis and Mining, 15(1), 82. https://doi.org/10.1007/s13278-025-01497-w
Fortuna, P., & Nunes, S. (2018a). A survey on automatic detection of hate speech in text. ACM Computing Surveys (CSUR), 51(4), 1–30. https://doi.org/10.1145/3232676
Fortuna, P., & Nunes, S. (2018b). A survey on automatic detection of hate speech in text. Acm Computing Surveys (Csur), 51(4), 1–30. https://doi.org/10.1145/3232676
Gaim, F., Song, H., Lee, H., Ko, C., Hwang, E. J., & Park, J. C. (2025). A Multi-Task Benchmark for Abusive Language Detection in Low-Resource Settings. ArXiv Preprint ArXiv:2505.12116.
Ghanem, R., & Erbay, H. (2020). Context-dependent model for spam detection on social networks. SN Applied Sciences, 2, 1–8. https://doi.org/10.1007/s42452-020-03374-x
Ghanem, R., & Erbay, H. (2023). Spam detection on social networks using deep contextualized word representation. Multimedia Tools and Applications, 82(3), 3697–3712. https://doi.org/10.1007/s11042-022-13397-8
Ghanem, R., Erbay, H., & Bakour, K. (2023). Contents-Based Spam Detection on Social Networks Using RoBERTa Embedding and Stacked BLSTM. SN Computer Science, 4(4), 380. https://doi.org/10.1007/s42979-023-01798-x
Gupta, V., Dixit, A., & Sethi, S. (2022). A Comparative Analysis of Sentence Embedding Techniques for Document Ranking. Journal of Web Engineering, 21(7), 2149–2185. https://doi.org/10.13052/jwe1540-9589.2177
Kapil, P., & Ekbal, A. (2025). A transformer based multi task learning approach to multimodal hate speech detection. Natural Language Processing Journal, 11, 100133. https://doi.org/10.1016/j.nlp.2025.100133
Kokkinos, C. M., Antoniadou, N., & Markos, A. (2014). Cyber-bullying: An investigation of the psychological profile of university student participants. Journal of Applied Developmental Psychology, 35(3), 204–214. https://doi.org/10.1016/j.appdev.2014.04.001
Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Lattanner, M. R. (2014). Bullying in the digital age: A critical review and meta-analysis of cyberbullying research among youth. Psychological Bulletin, 140(4), 1073–1137. https://doi.org/10.1037/a0035618
Kowalski, R. M., Toth, A., & Morgan, M. (2018). Bullying and cyberbullying in adulthood and the workplace. The Journal of Social Psychology, 158(1), 64–81. https://doi.org/10.1080/00224545.2017.1302402
Li, L., Xiao, L., Jin, W., Zhu, H., & Yang, G. (2018). Text Classification Based on Word2vec and Convolutional Neural Network. Neural Information Processing: 25th International Conference, ICONIP 2018, Siem Reap, Cambodia, December 13–16, 2018, Proceedings, Part V 25, 450–460. https://doi.org/10.1007/978-3-030-04221-9_40 Lumbantoruan, R., Siregar, R. U., Manik, I., Tambunan, N., & Simanjuntak, H. (2022). Analysis comparison of FastText and Word2vec for detecting offensive language. 2022 IEEE International Conference of Computer Science and Information Technology (ICOSNIKOM), 1–8. https://doi.org/10.1109/icosnikom56551.2022.10034886
Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representations in vector space. ArXiv Preprint ArXiv:1301.3781.
Mozafari, M., Farahbakhsh, R., & Crespi, N. (2022a). Cross-lingual few-shot hate speech and offensive language detection using meta learning. IEEE Access, 10, 14880–14896. https://doi.org/10.1109/access.2022.3147588
Mozafari, M., Farahbakhsh, R., & Crespi, N. (2022b). Cross-lingual few-shot hate speech and offensive language detection using meta learning. IEEE Access, 10, 14880–14896. https://doi.org/10.1109/access.2022.3147588
Nayel, H. A., & Shashirekha, H. L. (2019). DEEP at HASOC2019: A Machine Learning Framework for Hate Speech and Offensive Language Detection. FIRE (Working Notes), 336–343.
Oriola, O., & Kotzé, E. (2020). Evaluating machine learning techniques for detecting offensive and hate speech in South African tweets. IEEE Access, 8, 21496–21509. https://doi.org/10.1109/access.2020.2968173
Radha, N., & Swathika, R. (2025). SSN_IT_HATE@ LT-EDI-2025: Caste and Migration Hate Speech Detection. Proceedings of the 5th Conference on Language, Data and Knowledge: Fifth Workshop on Language Technology for Equality, Diversity, Inclusion, 84–89. https://doi.org/10.18653/v1/2024.ltedi-1.29
Reimers, N., & Gurevych, I. (2019). Sentence-bert: Sentence embeddings using siamese bert-networks. ArXiv Preprint ArXiv:1908.10084. https://doi.org/10.18653/v1/d19-1410
Rizwan, H., Shakeel, M. H., & Karim, A. (2020). Hate-speech and offensive language detection in roman Urdu. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2512–2522. https://doi.org/10.18653/v1/2020.emnlp-main.197
Roy, P. K., Bhawal, S., & Subalalitha, C. N. (2022). Hate speech and offensive language detection in Dravidian languages using deep ensemble framework. Computer Speech & Language, 75, 101386. https://doi.org/10.1016/j.csl.2022.101386
Sharif, O., Hossain, E., & Hoque, M. M. (2021). Nlp-cuet@ dravidianlangtech-eacl2021: Offensive language detection from multilingual code-mixed text using transformers. ArXiv Preprint ArXiv:2103.00455.
Sigurbergsson, G. I., & Derczynski, L. (2019). Offensive language and hate speech detection for Danish. ArXiv Preprint ArXiv:1908.04531.
Tawalbeh, S. K., Hammad, M., & Al-Smadi, M. (2020). KEIS@ JUST at SemEval-2020 Task 12: Identifying multilingual offensive tweets using weighted ensemble and fine-tuned BERT. ArXiv Preprint ArXiv:2005.07820. https://doi.org/10.18653/v1/2020.semeval-1.269
Waseem, Z., & Hovy, D. (2016). Hateful symbols or hateful people? predictive features for hate speech detection on twitter. Proceedings of the NAACL Student Research Workshop, 88–93. https://doi.org/10.18653/v1/n16-2013
Watanabe, H., Bouazizi, M., & Ohtsuki, T. (2018). Hate speech on twitter: A pragmatic approach to collect hateful and offensive expressions and perform hate speech detection. IEEE Access, 6, 13825–13835. https://doi.org/10.1109/access.2018.2806394
Wei, B., Li, J., Gupta, A., Umair, H., Vovor, A., & Durzynski, N. (2021). Offensive language and hate speech detection with deep learning and transfer learning. ArXiv Preprint ArXiv:2108.03305.
Zampieri, M., Malmasi, S., Nakov, P., Rosenthal, S., Farra, N., & Kumar, R. (2019). Predicting the type and target of offensive posts in social media. ArXiv Preprint ArXiv:1902.09666.
Zhang, Y., Hangya, V., & Fraser, A. (2025). LLM Sensitivity Challenges in Abusive Language Detection: Instruction-Tuned vs. Human Feedback. Proceedings of the 31st International Conference on Computational Linguistics, 2765–2780.
Zhang, Z., Robinson, D., & Tepper, J. (2018). Detecting hate speech on twitter using a convolution-gru based deep neural network. European Semantic Web Conference, 745–760. https://doi.org/10.1007/978-3-319-93417-4_48

Details

Primary Language

English

Subjects

Natural Language Processing

Journal Section

Research Article

Authors

Rezan Bakır ^*
0000-0002-4373-2231
Türkiye

Publication Date

March 3, 2026

Submission Date

April 18, 2025

Acceptance Date

December 5, 2025

Published in Issue

Year 2026 Volume: 29 Number: 1

DOI

https://doi.org/10.17780/ksujes.1679361

IZ

https://izlik.org/JA76LG63HW

Cite

RIS / Bibtex

APA

Bakır, R. (2026). TRI-EMBEDDINGS: A NOVEL APPROACH FOR DETECTING ABUSIVE LANGUAGE ON SOCIAL MEDIA. Kahramanmaraş Sütçü İmam Üniversitesi Mühendislik Bilimleri Dergisi, 29(1), 278-294. https://doi.org/10.17780/ksujes.1679361