CBLTwitter: CNN-BiLSTM derin öğrenme yöntemlerini kullanarak Twitter felaket tespiti

Yıl 2025, Cilt: 15 Sayı: 2, 563 - 576, 15.06.2025

Halit Çetiner , Hakan Yüksel

https://doi.org/10.17714/gumusfenbil.1653072

Öz

Sosyal medya platformlarından biri olan Twitter, herkesin düşünce ve fikirlerini çevrimiçi olarak dile getirmesini sağlayan güvenilir kaynaklardan biridir. Bu makalede, Twitter platformundaki tweet içeriklerindeki olası afet ya da felaket gibi olağanüstü durumlardaki metin içeriklerinin incelenmesi ve analizine odaklanılmıştır. Twitter platformundan alınan gerçek zamanlı bilgiler neticesinde olası felaket durumlarında insanlara yardımcı olmak ve acil durum ekiplerini otomatik olarak yönlendirme yapmak mümkündür. Belirtilen olası senaryoları gerçekleştirilmesine zemin hazırlayabilmek için binlerce ham metin içeriği içerisinden felaket ile ilgili içerikleri tespit ederek yüksek performans seviyesine sahip sınıflandırma gerçekleştirmek gerekmektedir. Bu makalede, felaketler hakkında karar verebilen ham tweet bilgileri içerisindeki yerel kalıpları ve bağlamsal bağımlılıkları yakalayabilen önemli değerlerinin ağırlık puanlarını artırarak sınıflandırma yapan CBLTwitter modeli önerilmiştir. Proposed CBLTwitter modeli, Twitter verilerinden felaket tahmin etmede Bidirectional Encoder Representations from Transformers (BERT) adlı bağlamsal kelime yerleştiricinin etkinliğini araştırmaktadır. Bunların yanısıra BERT sonuçlarının, Word2Vec ve Global Vectors for Word Representation (GloVe) adlı bağımsız kelime yerleştirme yöntemlerinden elde edilen sonuçlar ile karşılaştırılması yapılmaktadır. Sonuç olarak felaket tahmininde BERT kelime yerleştiricinin attention katmanlı Convolutional Neural Network (CNN) ve Bidirectional Long Short Term Memory (BiLSTM) mimarilerinden oluşan proposed CBLTwitter modeli literatür ile rekabet edebilir performans sonuçları sağlamıştır.

Anahtar Kelimeler

Dikkat mekanizması , BERT , BiLSTM , CNN , NLP , Twitter verisi

CBLTwitter: Twitter disaster detection analysis using CNN-BiLSTM deep learning methods

Öz

Twitter, one of the social media platforms, is one of the reliable sources that allows everyone to express their thoughts and ideas online. In this article, we focus on analysing and analysing the text content of tweets on the Twitter platform in extraordinary situations such as possible disasters or disasters. As a result of real-time information from the Twitter platform, it is possible to help people in possible disaster situations and automatically direct emergency teams. In order to prepare the ground for the realization of these possible scenarios, it is necessary to perform high performance classification by identifying disaster-related content from thousands of raw text content. In this paper, we propose a CBLTwitter model that classifies disasters by increasing the weight scores of their significant values that can capture local patterns and contextual dependencies in raw tweet information. The proposed CBLTwitter model investigates the effectiveness of a contextual word embedder called Bidirectional Encoder Representations from Transformers (BERT) in predicting disasters from Twitter data. In addition, BERT results are compared with the results obtained from independent word embedding methods called Word2Vec and Global Vectors for Word Representation (GloVe). As a result, the proposed CBLTwitter model of the BERT word embedder in disaster prediction, which consists of an attention-layer Convolutional Neural Network (CNN) and Bidirectional Long Short Term Memory (BiLSTM) architectures, provided performance results competitive with the literature.

Anahtar Kelimeler

Attention mechanism , BERT , BiLSTM , CNN , NLP , Twitter data

Kaynakça

• Acheampong, F. A., Nunoo-Mensah, H., & Chen, W. (2021). Transformer models for text-based emotion detection: a review of BERT-based approaches. Artificial Intelligence Review, 54(8), 5789–5829.
• Addison Howard, devrishi, Phil Culliton, Y. G. (2019, December 20). Natural language processing with disaster tweets. https://kaggle.com/competitions/nlp-getting-started/data.
• Al-Aidaroos, A. S., & Bamzahem, S. (2023). The impact of GloVe and Word2Vec word-embedding technologies on bug localization with convolutional neural network. International Journal of Science and Engineering Applications,12(1), 108-111.
• Alami, S., & Elbeqqali, O. (2015). Cybercrime profiling: text mining techniques to detect and predict criminal activities in microblog posts. 2015 10th International Conference on Intelligent Systems: Theories and Applications (SITA) (pp. 1–5), Rabat.
• Balakrishnan, V., Shi, Z., Law, C. L., Lim, R., Teh, L. L., Fan, Y., & Periasamy, J. (2022). A comprehensive analysis of transformer-deep neural network models in twitter disaster detection. In Mathematics, 10(24), 4664.
• Birjali, M., Kasri, M., & Beni-Hssane, A. (2021). A comprehensive survey on sentiment analysis: approaches, challenges and trends. Knowledge-Based Systems, 226, 107134.
• Biswas, R., & De, S. (2022). A comparative study on improving word embeddings beyond Word2Vec and GloVe. 2022 Seventh International Conference on Parallel, Distributed and Grid Computing (PDGC), (pp. 113–118), Solan.
• Çetiner, H. (2022). Multi-label text analysis with a CNN and LSTM based hybrid deep learning model. Adıyaman Üniversitesi Mühendislik Bilimleri Dergisi, 9(17), 447-457.
• Çetiner, H. (2023). Cataract disease classification from fundus images with transfer learning based deep learning model on two ocular disease datasets. Gümüşhane Üniversitesi Fen Bilimleri Dergisi, 13(2), 258-269.
• Çetiner, H. (2024). Fake news detection and classification with recurrent neural network based deep learning approaches. Osmaniye Korkut Ata Üniversitesi Fen Bilimleri Enstitüsü Dergisi, 7(3), 973–993.
• Çetiner, M. (2022). Sürdürülebilir moda ürünlerinin derin öğrenme yaklaşımı kullanarak analizi [Doktora Tezi, Süleyman Demirel Üniversitesi Sosyal Bilimler Enstitüsü].
• Cho, H. C., Okazaki, N., & Inui, K. (2013). Inducing context gazetteers from encyclopedic databases for named entity recognition. In Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013, Gold Coast, Australia, April 14-17, 2013, Proceedings, Part I 17 (pp. 378-389). Springer.
• Deb, S., & Chanda, A. K. (2022). Comparative analysis of contextual and context-free embeddings in disaster prediction from Twitter data. Machine Learning with Applications, 7, 100253.
• Devlin, J. (2018). Bert: pre-training of deep bidirectional transformers for language understanding. ArXiv Preprint ArXiv:1810.04805.
• Dharma, E. M., Gaol, F. L., Warnars, H., & Soewito, B. (2022). The accuracy comparison among word2vec, glove, and fasttext towards convolution neural network (CNN) text classification. J Theor Appl Inf Technol, 100(2), 31.
• Eight, F. (2019, February 21). Twitter Airline Sentiment. https://www.kaggle.com/datasets/crowdflower/twitter-airline-sentiment.
• Feng, X., Angkawisittpan, N., & Yang, X. (2024). A CNN-BiLSTM algorithm for weibo emotion classification with attention mechanism. Mathematical Models in Engineering, 10(2), 87-97.
• Huang, Q., Chen, R., Zheng, X., & Dong, Z. (2017). Deep sentiment representation based on CNN and LSTM. 2017 International Conference on Green Informatics (ICGI) (pp. 30–33), Fuzhou.
• Khan, S., & Yairi, T. (2018). A review on the application of deep learning in system health management. Mechanical Systems and Signal Processing, 107, 241–265.
• Khatua, A., Khatua, A., & Cambria, E. (2019). A tale of two epidemics: contextual Word2Vec for classifying twitter streams during outbreaks. Information Processing & Management, 56(1), 247–257.
• Kim, S., & Lee, S. P. (2023). A BiLSTM–transformer and 2D CNN architecture for emotion recognition from speech. Electronics, 12(19), 4034.
• Kishwar, A., & Zafar, A. (2023). Fake news detection on Pakistani news using machine learning and deep learning. Expert Systems with Applications, 211, 118558.
• Kowsher, M., Tahabilder, A., Islam Sanjid, M. Z., Prottasha, N. J., Uddin, M. S., Hossain, M. A., & Kader Jilani, M. A. (2021). LSTM-ANN & BiLSTM-ANN: Hybrid deep learning models for enhanced classification accuracy. Procedia Computer Science, 193, 131–140.
• Lin, K., & Pomerleano, D. (2011). Global matrix factorizations. Mathematical Research Letters, 20.
• Mahajan, P., Raghuwanshi, P., Setia, H., & Randhawa, P. (2024). A multi-model approach for disaster-related tweets: a comparative study of machine learning and neural network models. Journal of Computers, Mechanical and Management, 3(2), 19-24.
• Manthena, S. P. (2023). Leveraging tweets for rapid disaster response using BERT-BiLSTM-CNN model [Master of Science, San Jose State University Department of Computer Science].
• Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781.
• Neethu, M. S., & Rajasree, R. (2013). Sentiment analysis in twitter using machine learning techniques. 2013 Fourth International Conference on Computing, Communications and Networking Technologies (ICCCNT) (pp. 1–5), Tiruchengode.
• Pennington, J., Socher, R., & Manning, C. D. (2014, October). Glove: global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP) (pp. 1532-1543), Doha.
• Priya, C. S. R., & Deepalakshmi, P. (2023). Sentiment analysis from unstructured hotel reviews data in social network using deep learning techniques. International Journal of Information Technology, 15(7), 3563–3574.
• R., M., S., M., OS, N., & E., T. (2023). An enhanced framework for disaster-related tweet classification using machine learning techniques. 2023 International Conference on Inventive Computation Technologies (ICICT) (pp. 108–111), Nepal.
• Rajesh, A., & Hiwarkar, T. (2023). Sentiment analysis from textual data using multiple channels deep learning models. Journal of Electrical Systems and Information Technology, 10(1), 56.
• Rakshit, P., & Sarkar, A. (2025). A supervised deep learning-based sentiment analysis by the implementation of Word2Vec and GloVe embedding techniques. Multimedia Tools and Applications, 84, 979-1012.
• Rosenthal, S., Farra, N., & Nakov, P. (2017). SemEval-2017 task 4: Sentiment analysis in twitter. In S. Bethard, M. Carpuat, M. Apidianaki, S. M. Mohammad, D. Cer, D. Jurgens (eds.), Proceedings of the 11th international workshop on semantic evaluation (SemEval-2017) (pp. 502–518). Vancouver: Association for Computational Linguistics.
• Sadr, H., & Nazari Soleimandarabi, M. (2022). A CNN-TL: attention-based convolutional neural network coupling with transfer learning and contextualized word representation for enhancing the performance of sentiment classification. The Journal of Supercomputing, 78(7), 10149–10175.
• Semary, N. A., Ahmed, W., Amin, K., Pławiak, P., & Hammad, M. (2023). Improving sentiment classification using a RoBERTa-based hybrid model. Frontiers in Human Neuroscience, 17.
• Sermanet, P., Eigen, D., Zhang, X., Mathieu, M., Fergus, R., & LeCun, Y. (2013). Overfeat: integrated recognition, localization and detection using convolutional networks. ArXiv Preprint ArXiv:1312.6229.
• Shaik, T., Tao, X., Dann, C., Xie, H., Li, Y., & Galligan, L. (2023). Sentiment analysis and opinion mining on educational data: a survey. Natural Language Processing Journal, 2, 100003.
• Sitaula, C., & Shahi, T. B. (2024). Multi-channel CNN to classify nepali COVID-19 related tweets using hybrid features. Journal of Ambient Intelligence and Humanized Computing, 15(3), 2047–2056.
• Sitender, Sangeeta, Sushma, N. S., & Sharma, S. K. (2023). Effect of GloVe, Word2Vec and FastText embedding on english and hindi neural machine translation systems. In Proceedings of Data Analytics and Management: ICDAM 2022 (pp. 433-447). Singapore: Springer Nature Singapore.
• Song, G., & Huang, D. (2021). A sentiment-aware contextual model for real-time disaster prediction using twitter data. Future Internet, 13(7), 163.
• Sukhbaatar, S., Weston, J., & Fergus, R. (2015). End-to-end memory networks. Advances in Neural Information Processing Systems, 28.
• Tam, S., Said, R. Ben, & Tanriöver, Ö. Ö. (2021). A ConvBiLSTM deep learning model-based approach for twitter sentiment classification. IEEE Access, 9, 41283–41293.
• Tan, K. L., Lee, C. P., Anbananthen, K. S. M., & Lim, K. M. (2022). RoBERTa-LSTM: a hybrid model for sentiment analysis with transformer and recurrent neural network. IEEE Access, 10, 21517–21525.
• Vadivukarassi, M., Puviarasan, N., & Aruna, P. (2018). An exploration of airline sentimental tweets with different classification model. International Journal for Research in Engineering Application & Management, 4(2).
• Vaswani, A. (2017). Attention is all you need. ArXiv Preprint ArXiv:1706.03762.
• Wankhade, M., Annavarapu, C. S. R., & Abraham, A. (2024). CBMAFM: CNN-BiLSTM multi-attention fusion mechanism for sentiment classification. Multimedia Tools and Applications, 83(17), 51755–51786.
• Yang, Y., & Li, S. (2024). Entity overlapping relation extracting algorithm based on CNN and BERT. IEEE Access, 1.
• Yeboah, P. N., & Baz Musah, H. B. (2022). NLP technique for malware detection using 1D CNN fusion model. Security and Communication Networks, 2022(1), 2957203.
• Zhao, J., Liu, K., & Xu, L. (2016). Sentiment analysis: mining opinions, sentiments, and emotions. Cambridge University Press.