Sentiment analysis on social-media textual data using hybrid transformer and sequence models

Verma, Harihar Nath; Malik, Sanjay Kumar; Jain, Vanita

doi:10.47974/JDMSC-2633

Open Access Research Article

Sentiment analysis on social-media textual data using hybrid transformer and sequence models

Harihar Nath Vermahnvermagwl@gmail.comUniversity School of Information, Communication and Technology Guru Gobind Singh Indraprastha UniversityUniversity School of Information, Communication and Technology Guru Gobind Singh Indraprastha UniversityDwarka, Delhi, 110078, IndiaView full profile → , Sanjay Kumar Malikskmalik@ipu.ac.inUniversity School of Information, Communication and Technology Guru Gobind Singh Indraprastha UniversityUniversity School of Information, Communication and Technology Guru Gobind Singh Indraprastha UniversityDwarka, Delhi, 110078, IndiaView full profile → , ^*Vanita JainCorresponding authorvanitajain@fot.du.ac.inDepartment of Electronics and Communication Engineering Faculty of Technology University of DelhiDepartment of Electronics & Communication Engineering Faculty of Technology University of Delhi Maharishi Kanad BhawanDelhi, 110007, IndiaView full profile →

* Corresponding author · click or hover a name for details

pp. 375–396Vol. 29Issue 1January 2026DOI: 10.47974/JDMSC-2633 Crossmark XML

Received:: 15 Jan 2025
Published Online:: 15 Jan 2026
Article type:: Research Article
Language:: EN
Article no.:: JDMSC-2633
Pages:: 375–396

Abstract

Sentiment analysis has become an invaluable exercise on social media addressing the understanding of opinion and consumer behavior and trends. The system of informality and context richness against platforms, such as Twitter, raises serious issues to current sentiment classification models. It is hypothesized that integration of the contextual feature representation capacity of embeddings in transformers with the sequential learning ability of BiLSTM networks in DitilBERT-BiLSTM can help to improve accuracy in sentiment classification. The Sentiment140 dataset is used as a training and testing set with 1.6 million small tweets among which there are equal numbers of positive and negative tweets. Experiments demonstrate that the proposed model obtains superior performance compared to traditional deep learning architectures, 81% accuracy, 82% precision, recall of 80% and F1-score of 81%. DistilBERT-BiLSTM shows good generalization and learns efficiently using less computer power than full-scale transformer models, indicating that it is more suitable to applications of real-world social media sentiment analysis. This study points out the success of hybrid deep learning networks in modelling both global semantic structure and local sequential relations as a strong remedy to resolving opinion mining in dynamic and noisy digital situation.

Keywords

Sentiments Tweets Social media text DistilBERT BiLSTM Hybrid deep learning model Transformer-based embeddings Natural language processing Sentiment140 dataset Opinion mining

Subject Classifications

68T0768T50

References

[1] L. Chaudhary, N. Girdhar, D. Sharma, J. Andreu-perez, and A. Doucet, “A Review of Deep Learning Models for Twitter Sentiment Analysis: Challenges and Opportunities”, Journal of Latex Class Files, vol. 14, no. 8, pp. 1–30 (2015).
[2] H. Pal and B. Bhushan, “Sentiment Analysis on Twitter Dataset using Voting Classifier”, International Conference on Electrical Electronics and Computing Technologies (ICEECT), India, pp. 1–6 (Aug. 2024), doi: 10.1109/ICEECT61758.2024.10739316.
[3] M. R. Baker, and A. Utku, “Unraveling user perceptions and biases: A comparative study of ML and DL models for exploring twitter sentiments towards ChatGPT”, Journal of Engineering Research, vol. 13, no. 2, pp 1658–1665 (2025), doi: 10.1016/j.jer.2023.11.023.
[4] A. Jain, and V. Jain, “Sentiment classification of twitter data belonging to renewable energy using machine learning”, Journal of Information and Optimization Sciences, vol. 40, no. 2, pp. 521–533 (Mar. 2019). doi: 10.1080/02522667.2019.1582873.
[5] K. L. Tan, C. P. Lee, and K. M. Lim, “A Survey of Sentiment Analysis: Approaches, Datasets, and Future Research”, Applied Sciences, vol. 13, no. 7, pp 1–21 (2023), doi: 10.3390/app13074550.
[6] B. Pang and L. Lee, “Opinion Mining and Sentiment Analysis”, Foundations and Trends in Information Retrieval, vol. 2, no. 1–2, pp. 1–135 (2008), doi: 10.1561/1500000011.
[7] J. Devlin, M. Chang, K. Lee, and K. Tautanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding”, Computation and Language (May 2019). doi: 10.48550/arXiv.1810.04805.
[8] A. A. Khan, S. K. Malik, and V. Jain, “Semantic search framework over knowledge bases using embeddings-based similarity”, Journal of Discrete Mathematical Sciences & Cryptography, vol. 27, No. 6, pp. 1963–1975 (Sep. 2024). doi: 10.47974/JDMSC-2047.
[9] https://www.kaggle.com/datasets/kazanova/sentiment140.
[10] Sanh, L. Debut, J. Chaumond, and T. Wolf, “DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter”, Computation and Language, pp 1–5 (Mar. 2020). doi: 10.48550/arXiv.1910.01108.
[11] V. Jain, P. S. Lamba, B. Singh, N. Namboothiri, and S. Dhall, “Facial expression recognition using feature level fusion”, Journal of Discrete Mathematical Sciences and Cryptography, vol. 22, no. 2, pp. 337–350 (Mar. 2019). doi: : 10.1080/09720529.2019.1582866.
[12] J. Wang, J. Wei, and F. Tian, “A comparative study of machine learning models for sentiment analysis of transboundary rivers news media articles”, Soft Computing, vol. 28, no. 23, pp. 13331–13347 (Dec. 2024). doi: 10.1007/s00500-024-10357-2.
[13] Golubev, N. Rusnachenko, and N. Loukachevitch, “RuSentNE-2023: Evaluating Entity-Oriented Sentiment Analysis on Russian News Texts” (May 2023). doi: 10.28995/2075-7182-2022-20-XX-XX.
[14] J. Tang, “Leveraging Transformer Neural Networks for Enhanced Sentiment Analysis on Online Platform Comments”, 4th International Conference on Machine Learning and Computer Application (ICMLCA 2023), ACM, pp. 256–260 (Nov. 2023). doi: 10.1145/3650215.3650260.
[15] A. Sharma, and R. Bhalla, “Detecting Hate Speech for Hindi-English Code-Mix Text Data Using Dual Contrastive Learning”, Sixth International Conference on Futuristic Trends in Networks and Computing Technologies (FTNCT06), vol. 259, pp 35–43 (2025). doi: 10.1016/j.procs.2025.03.304.
[16] A. Ilham, A. Bustamin, A. A. Prayogi, Safina, and A. R. Rahmika, “Development of Sentiment Analysis and Sarcasm Detection Systems for Social Network Users”, 4th EPI International Conference on Science and Engineering, EICSE 2020 - Gowa, Indonesia, AIP Conference Proceedings, vol. 2543 (Nov. 2022). doi: 10.1063/5.0094737.
[17] https://www.kaggle.com/datasets/tegzes/isarcasm.
[18] T. Handoyo, Hidayaturrahman, C. J. Setiadi, and D. Suhartono, “Sarcasm Detection in Twitter - Performance Impact While Using Data Augmentation: Word Embeddings”, International Journal of Fuzzy Logic and Intelligent Systems, vol. 22, no. 4, pp. 401–413 (Dec. 2022), doi: 10.5391/IJFIS.2022.22.4.401.
[19] N. Malali, “AI-Powered Data Preprocessing and Transformation Platform for Autonomous Data Cleaning, Advanced Fea”, Patent no. 202521035175 (Apr. 2025).
[20] T. UcKan, “A hybrid model for extractive summarization: Leveraging graph entropy to improve large language model performance”, Ain Shams Engineering Journal, vol. 16, no. 5, pp 1–15 (Mar. 2025). doi: 10.1016/j.asej.2025.103348.
[21] R. Setiyawan and Z. Mustofa, “Comparison of the performance of naive bayes and support vector machine in sirekap sentiment analysis with the lexicon-based approach”, Journal of Soft Computing Exploration, vol. 5, no. 2, pp. 122–132 (June 2024), doi: 10.52465/joscex.v5i2.367.
[22] S. Adhikary, “A Machine Learning Approach for COVID-19 Tweet Classification”, Forum for Information Retrieval Evaluation, CEUR Workshop Proceedings, vol. 3395, pp. 349–353 (2022).
[23] O. Mairaj, and S. U. R. Khan, “Unveiling temporal patterns in information for improved rumor detection”, Social Network Analysis and Mining, vol. 15, no. 13, pp 1–14 (Mar. 2025). doi: 10.1007/s13278-025-01421-2.

Views: 100Downloads: 10Citations: 0