AN IMPROVED HYBRID DEEP LEARNING MODEL FOR SENTIMENT CLASSIFICATION, OPINION CLUSTERING, AND TEMPORAL REASONING IN PRODUCT REVIEWS

Author

Pooja Dwivedi, Dr. Sanjay Kumar

Abstract

 Sentiment classification models built on transformer encoders have substantially advanced opinion mining, yet they continue to struggle with domain adaptation, sarcasm, multilingual and imbalanced data, and, more fundamentally, with interpretability. At the same time, clustering techniques used to group customers or opinions remain largely static and rule-based, and existing sentiment pipelines almost entirely ignore the temporal evolution of consumer opinion. This paper proposes an integrated model comprising three coupled components: (i) an improved sentiment classifier that fuses transformer encoders (RoBERTa/T5/DeBERTa) with bidirectional LSTM layers and attention-based signal fusion, combined with SMOTE and focal loss to address class imbalance; (ii) a novel sentiment-aware clustering technique that constructs an opinion-similarity graph and applies fuzzy c-means clustering optimized by genetic algorithms/particle swarm optimization to reveal latent customer segments; and (iii) a temporal reasoning module that uses LSTM/GRU-based time-series modelling with temporal embeddings and multi-scale attention fusion to predict evolving customer requirements. Together these modules are intended to improve recall, precision, and interpretability of opinion mining while enabling dynamic, forward-looking decision support for e-commerce and social-media platforms. The paper describes the architecture, implementation steps, and evaluation protocol (accuracy, F1-score, NMI, Silhouette coefficient, MAE, RMSE) proposed for validating the integrated model.

Keywords

sentiment classification; opinion mining; aspect-based sentiment analysis; fuzzy c-means clustering; temporal reasoning; BiLSTM; attention mechanism; transformer; class imbalance; product reviews

Full Text:

Download Paper PDF

References

[1] Zhang, Z., Zou, Y., & Gan, C. (2018). Textual sentiment analysis via three different attention convolutional neural networks and cross-modality consistent regression. Neurocomputing, 275, 1407-1415.

[2] Abid, F., Alam, M., Yasir, M., & Li, C. (2019). Sentiment analysis through recurrent variants latterly on convolutional neural network of Twitter. Future Generation Computer Systems, 95, 292-308.

[3] Song, M., Park, H., & Shin, K. S. (2019). Attention-based long short-term memory network using sentiment lexicon embedding for aspect-level sentiment analysis in Korean. Information Processing & Management, 56(3), 637-653.

[4] Chen, F., & Huang, Y. (2019). Knowledge-enhanced neural networks for sentiment analysis of Chinese reviews. Neurocomputing, 368, 51-58.

[5] Rasappan, P., et al. (2024). Transforming sentiment analysis for e-commerce product reviews: Hybrid deep learning model with an innovative term weighting and feature selection. Information Processing & Management, 61.

[6] Wang, D., Li, J., Xu, K., & Wu, Y. (2017). Sentiment community detection: exploring sentiments and relationships in social networks. Electronic Commerce Research, 17(1), 103-132.

[7] Liu, Y., Bi, J. W., & Fan, Z. P. (2017). Ranking products through online reviews: A method based on sentiment analysis technique and intuitionistic fuzzy set theory. Information Fusion, 36, 149-161.

[8] Preethi, P. G., & Uma, V. (2015). Temporal sentiment analysis and causal rules extraction from tweets for event prediction. Procedia Computer Science, 48, 84-89.

[9] Ibrahim, N. F., & Wang, X. (2019). Decoding the sentiment dynamics of online retailing customers: Time series analysis of social media. Computers in Human Behavior, 96, 32-45.

[10] Torres, E. N., Wei, W., Hua, N., & Chen, P. J. (2019). Customer emotions minute by minute: How guests experience different emotions within the same service environment. International Journal of Hospitality Management, 77, 128-138.

[11] Tang, F., Fu, L., Yao, B., & Xu, W. (2019). Aspect based fine-grained sentiment analysis for online reviews. Information Sciences, 488, 190-204.

[12] Ray, P., & Chakrabarti, A. (2020). A mixed approach of deep learning method and rule-based method to improve aspect level sentiment analysis. Applied Computing and Informatics.

[13] Guo, X. (2024). Sentiment analysis based on RoBERTa for Amazon review: An empirical study on decision making. arXiv preprint arXiv:2411.00796.

[14] Kayed, M., Díaz-Redondo, R. P., & Mabrouk, A. (2023). Deep learning-based sentiment classification: A comparative survey. arXiv preprint arXiv:2312.17253.

[15] Xia, R., Xu, F., Yu, J., Qi, Y., & Cambria, E. (2016). Polarity shift detection, elimination and ensemble: A three-stage model for document-level sentiment analysis. Information Processing & Management, 52(1), 36-45.