ReviewSense AI Sentiment Analysis of Product Reviews Using Deep Learning

This paper set out ReviewSense AI, a hybrid CNN–BiLSTM–Attention architecture for sentiment classification of product reviews built on concatenated GloVe and BERT embeddings, along with the system design, testing plan, and deployment considerations needed to run it as a real service rather than an offline experiment. Tested against seven trained baselines and several existing sentiment-analysis tools on a combined Amazon/Yelp corpus, the model reached 97.6% accuracy and a 97.4% macro F1-score, and its attention weights and aspect-tagging module give it a level of interpretability that black-box transformer classifiers or hosted cloud APIs simply don't offer. That said, the research-gap analysis in Section III and the limitations in Section XIII make clear that beating a single benchmark isn't really the most pressing open problem in this space anymore — cross-domain robustness, sarcasm handling, multilingual coverage, and resilience to manipulated reviews are the directions that matter more going forward.

[1] B. Pang and L. Lee, “Opinion Mining and Sentiment Analysis,” Foundations and Trends in Information Retrieval, vol. 2, nos. 1-2, pp. 1-135, 2008. [2] H.-Y. Chen and Y.-S. Chen, “Deep Learning for Sentiment Analysis on Google Play Consumer Review,” in Proc. 2017 IEEE Int. Conf. on Applied System Innovation, 2017, pp. 1-4. [3] S. Gupta and A. Verma, “Sentiment Analysis of Customer Product Reviews Using Machine Learning,” in Proc. 2018 IEEE Int. Conf. on Intelligent Sustainable Systems, 2018. [4] C. J. Hutto and E. Gilbert, “VADER: A Parsimonious Rule-Based Model for Sentiment Analysis of Social Media Text,” in Proc. 8th Int. AAAI Conf. on Weblogs and Social Media (ICWSM), 2014. [5] Y. Kim, “Convolutional Neural Networks for Sentence Classification,” in Proc. EMNLP, 2014, pp. 1746-1751. [6] S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory,” Neural Computation, vol. 9, no. 8, pp. 1735-1780, 1997. [7] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” in Proc. NAACL-HLT, 2019, pp. 4171-4186. [8] A. Vaswani et al., “Attention Is All You Need,” in Proc. NeurIPS, 2017, pp. 5998-6008. [9] W. Zhang, X. Li, Y. Deng, L. Bing, and W. Lam, “A Survey on Aspect-Based Sentiment Analysis: Tasks, Methods, and Challenges,” IEEE Trans. Knowledge and Data Engineering, vol. 35, no. 11, pp. 11019-11038, 2023. [10] J. Pennington, R. Socher, and C. D. Manning, “GloVe: Global Vectors for Word Representation,” in Proc. EMNLP, 2014, pp. 1532-1543.