A Transformer-Based Model for Sentiment Analysis on Big Data Platforms
Keywords:
CNN, RCNN, Sentiment analysis, big data, adversarial learning, ParserBert, FaberBert, multilingual BERTAbstract
This study addresses the challenges of sentiment classification arising from the vastness of textual data by introducing a three-phase adversarial fine-tuning framework. In this framework, a base model is trained using a transformer encoder, multi-head attention, and a BiLSTM layer on preprocessed data, followed by the creation of a systematic generator that applies fourteen controlled perturbations at varying intensities to the adversarial datasets. In the base article, stage-wise training was conducted sequentially on preprocessed data and on different levels of adversarial datasets, demonstrating significant improvements particularly on a noisy dataset (level 3). Overall, this approach effectively enhances the robustness and reliability of sentiment classifiers in big data contexts with corrupted text. In the proposed method, by incorporating an optimized RCCN network and multi-agent SVM-based clustering, factors were re-clustered, resulting in the proposed algorithm achieving an accuracy rate of 97.86%.
Downloads
References
Ahmmed Babu, T., Aftahee, S., Kalimullah Ratu, M., & Hossain, J. M. H. M. (2025). A Transformer-Driven Approach to Political Sentiment Analysis in Tamil X (Twitter) Comments. https://doi.org/10.18653/v1/2025.dravidianlangtech-1.93
Amin, R., Gantassi, R., Ahmed, N., Alshehri, A. H., Alsubaei, F. S., & Frnda, J. (2024). A hybrid approach for adversarial attack detection based on sentiment analysis model using machine learning. Engineering Science and Technology, an International Journal, 58SP - 101829. https://doi.org/10.1016/j.jestch.2024.101829
Ashok Kumar Durairaj, A., & Chinnalagu, A. (2021). Transformer based Contextual Model for Sentiment Analysis of Customer Reviews: A Fine-tuned BERT. (IJACSA) International Journal of Advanced Computer Science and Applications, 12(11). https://doi.org/10.14569/IJACSA.2021.0121153
Davar, O., & Eftekhari, M. (2024). A Hybrid Method of Self-Supervised Graph Embedding, Siamese Networks, and Transformers for Sentiment Analysis in Persian Language. https://doi.org/10.1109/AISP61396.2024.10475270
Jafarian, H., Taghavi, A. H., Javaheri, A., & Rawassizadeh, R. (2021). Exploiting BERT to improve aspect-based sentiment analysis performance on Persian language. https://doi.org/10.1109/ICWR51868.2021.9443131
Karami, B., Bakouie, F., & Gharibzadeh, S. (2023). A transformer-based deep learning model for Persian moral sentiment analysis. Journal of Information Science, 01655515231188344. https://doi.org/10.1177/01655515231188344
Paulraj, D., Ezhumalai, P., & Prakash, M. (2024). A Deep Learning Modified Neural Network (DLMNN) based proficient sentiment analysis technique on Twitter data. Journal of Experimental & Theoretical Artificial Intelligence, 36(3).
Petratos, P., & Giannoula, M. (2024). Transformer-based AI for Sentiment Analysis in Marketing. https://doi.org/10.1145/3690771.3690795
Saxena, A., Reddy, H., & Saxena, P. (2022). Introduction to sentiment analysis covering basics, tools, evaluation metrics, challenges, and applications. In (pp. 249-277). https://doi.org/10.1007/978-981-16-3398-0_12
Vakili, Y. Z., Fallah, A., & Zakeri, S. (2024). Enhancing Sentiment Analysis of Persian Tweets: A Transformer-Based Approach. https://doi.org/10.1109/ICWR61162.2024.10533353
Zardak, S. R., Rasekh, A. H., & Bashkari, M. S. (2023). Persian Text Sentiment Analysis Based on BERT and Neural Networks. Iranian Journal of Science and Technology, Transactions of Electrical Engineering, 47(4), 1623-1634. https://doi.org/10.1007/s40998-023-00626-5
Downloads
Published
Submitted
Revised
Accepted
Issue
Section
License
Copyright (c) 2024 Mohammad Reza Noroozi, Ali Moeini (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
