Online product reviews have become an invaluable resource for consumers seeking information to make informed purchasing decisions. Sentiment analysis, a branch of Natural Language Processing (NLP), plays a crucial role in automatically extracting sentiments or opinions from these reviews. This research paper presents a comprehensive study on sentiment analysis in online product reviews, specifically focusing on Amazon product reviews. The study employs advanced machine learning techniques, including feature engineering and deep learning models such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Bidirectional Encoder Representations from Transformers (BERT), to accurately classify sentiment expressed in customer reviews. The methodology encompasses data collection from Amazon reviews, comprehensive preprocessing techniques including tokenization, stop-word removal, and feature extraction, followed by model training and evaluation. The BERT model demonstrates superior performance with 89% accuracy, 88% precision, 89% recall, and 88% F1-score, significantly outperforming traditional machine learning approaches such as Logistic Regression (83.1% accuracy) and Decision Trees (75% accuracy). This research provides valuable insights for e-commerce businesses to understand customer opinions better and make data-driven decisions regarding product improvements, marketing strategies, and customer satisfaction enhancement.

This research has presented a comprehensive study on sentiment analysis in online product reviews using advanced machine learning techniques, with specific focus on Amazon product reviews across multiple categories. The study successfully developed and evaluated a robust sentiment analysis framework that combines sophisticated preprocessing techniques, comprehensive feature engineering, and state-of-the-art deep learning models to achieve accurate sentiment classification. The BERT model demonstrated exceptional performance, achieving 89% accuracy, 88% precision, 89% recall, and 88% F1-score, significantly outperforming traditional machine learning approaches such as Logistic Regression (83.1% accuracy) and Decision Trees (75% accuracy). These results validate the effectiveness of transformer-based architectures and contextual embeddings for capturing the nuanced nature of sentiment expression in customer reviews. The superior performance of BERT stems from its bidirectional context modeling, attention mechanisms, and ability to capture long-term dependencies in text. The comprehensive methodology developed in this research, encompassing data collection, preprocessing, feature engineering, and model training, provides a systematic framework that can be applied to sentiment analysis tasks in various domains beyond e-commerce. The preprocessing pipeline effectively handles noisy text data, informal language, and diverse linguistic patterns common in online reviews. The feature engineering approach successfully combines multiple representation techniques to capture lexical, syntactic, and semantic aspects of sentiment expression. The practical implications of this research extend to multiple stakeholders in the e-commerce ecosystem. Businesses can leverage automated sentiment analysis to gain real-time insights into customer opinions, enabling data-driven decisions regarding product improvements, marketing strategies, and customer service enhancements. The ability to process thousands of reviews efficiently and accurately provides scalability that manual analysis cannot match. Consumers benefit from sentiment analysis through improved product recommendations and summarization of collective opinions. The research also contributes to the broader field of natural language processing by demonstrating effective techniques for handling class imbalance, neutral sentiment classification, and domain- specific language patterns. The comparative analysis of different models and feature representations provides valuable insights for researchers and practitioners developing sentiment analysis systems. Despite the impressive results, the study acknowledges certain limitations and challenges. Neutral sentiment classification remains difficult, with higher error rates compared to positive and negative categories. The model occasionally struggles with sarcasm, irony, and highly technical language. Very short or extremely long reviews present classification challenges due to insufficient or excessive contextual information. Computational requirements of the BERT model, while justified by performance gains, may limit deployment in resource-constrained environments. The findings of this research have significant implications for the future of sentiment analysis in e-commerce and beyond. The demonstrated effectiveness of deep learning approaches, particularly BERT, suggests that continued advancement in transformer architectures and pre- training techniques will drive further improvements. The framework developed in this study provides a foundation for extending sentiment analysis to multilingual contexts, multimodal data (incorporating images and videos), and fine-grained aspect-based sentiment analysis. In conclusion, this research successfully achieved its objectives of developing an accurate and robust sentiment analysis system for online product reviews, demonstrating the superiority of advanced deep learning techniques, and providing practical insights for e-commerce applications. The comprehensive methodology, rigorous evaluation, and detailed analysis contribute valuable knowledge to both the academic understanding of sentiment analysis and practical implementation of customer opinion mining systems.

[1]T. U. Haque, N. N. Saber, and F. M. Shah, “Sentiment analysis on large scale Amazon product reviews,” in 2018 IEEE International Conference on Innovative Research and Development (ICIRD), pp. 1-6, 2018. [2]X. Fang and J. Zhan, “Sentiment analysis using product review data,” Journal of Big Data, vol. 2, no. 1, pp. 1-14, 2015. [3]R. S. Kumar and R. Saviour Devaraj, “Sentiment analysis using semantic-based feature selection approach utilizing SentiWordNet,” International Journal of Advanced Computer Science and Applications, vol. 12, no. 9, pp. 123-134, 2021. [4]B. Kaur and N. Kumari, “A hybrid approach to sentiment analysis of technical article reviews,” International Journal of Education and Management Engineering, vol. 6, no. 6, pp. 1- 9, 2016. [5]C. Messaoudi, Z. Guessoum, and L. Ben Romdhane, “Opinion mining and sentiment analysis: A survey,” Expert Systems with Applications, vol. 185, pp. 115-129, 2022. [6]Q. T. Ghous, “Sentiment analysis of Amazon customer product reviews: A review,” International Journal of Scientific Research Engineering & Development, vol. 4, pp. 564-595, 2021. [7]V. Umarani, A. Julian, and J. Deepa, “Sentiment analysis using various machine learning and deep learning techniques,” Journal of the Nigerian Society of Physical Sciences, vol. 3, pp. 385-394, 2021. [8]B. Saberi, “Opinion mining or sentiment analysis: A survey,” International Journal of Advanced Research in Computer Science, vol. 8, no. 5, pp. 1234-1241, 2017. [9]S. Tammina and S. Annareddy, “Sentiment analysis on customer reviews using a convolutional neural network,” in 2020 International Conference on Computer Communication and Informatics (ICCCI), pp. 1-6, 2020. [10]M. Kabir, M. M. J. Kabir, S. Xu, and B. Badhon, “An empirical research on sentiment analysis using machine learning approaches,” International Journal of Computer Applications, vol. 182, no. 14, pp. 15-19, 2021. [11]E. Maks, “A lexicon model for deep sentiment analysis and opinion mining applications,” Decision Support Systems, vol. 53, no. 4, pp. 680-688, 2012. [12]M. Wahyudi and D. A. Kristiyanti, “Sentiment analysis of smartphone product review using support vector machine algorithm-based particle swarm optimization,” Journal of Theoretical and Applied Information Technology, vol. 91, no. 1, pp. 189-201, 2016. [13]E. Maks, “A lexicon model for deep sentiment analysis and opinion mining applications,” Computational Linguistics, vol. 38, no. 2, pp. 245-267, 2012. [14]N. Shrestha and F. Nasoz, “Deep learning sentiment analysis of Amazon.com reviews and ratings,” International Journal of Soft Computing and Artificial Intelligence, vol. 8, no. 1, pp. 01-09, 2019. [15]R. S. Jagdale, V. S. Shirsat, and S. N. Deshmukh, “Sentiment analysis on product reviews using machine learning techniques,” in Advances in Intelligent Systems and Computing, vol. 810, pp. 639-647, 2019. [16]B. Saberi, “Opinion mining or sentiment analysis: A survey of techniques,” International Journal of Computer Applications, vol. 168, no. 9, pp. 23-29, 2017. [17]Q. Yu, H. Zhao, and Z. Wang, “Attention-based bidirectional gated recurrent unit neural networks for sentiment analysis,” in ACM International Conference Proceeding Series, pp. 48-52, 2019. [18]F. Tang, L. Fu, B. Yao, and W. Xu, “Aspect based fine- grained sentiment analysis for online reviews,” Information Sciences, vol. 488, pp. 190-204, 2019. [19]W. Tan, X. Wang, and X. Xu, “Sentiment analysis for Amazon reviews using RNN-GRU models,” Research Report, pp. 1-9, March 2019. [20]R. S. Kumar and R. Saviour Devaraj, “Feature selection for sentiment analysis using SentiWordNet,” IEEE Access, vol. 9, pp. 45123-45138, 2021. [21]S. Alaparthi and M. Mishra, “BERT: A sentiment analysis odyssey,” Journal of Marketing Analytics, vol. 9, pp. 118-126, 2021. [22] C. Sun, X. Qiu, Y. Xu, and X. Huang, “How to fine-tune BERT for text classification,” in Lecture Notes in Computer Science, vol. 11856, pp. 194-206, 2019. [23]Q. T. Ghous, “Sentiment analysis of Amazon customer product reviews,” International Journal of Scientific Research Engineering & Development, vol. 4, no. 6, pp. 564-595, 2021. [24]C. Messaoudi, Z. Guessoum, and L. Ben Romdhane, “Opinion mining and sentiment analysis challenges: A comprehensive survey,” Expert Systems with Applications, vol. 203, pp. 117-135, 2022. [25]A. Abayomi-Alli, O. Abayomi-Alli, S. Misra, and L. Fernandez-Sanz, “Sentiment analysis of social media data using machine learning,” Computer Networks, vol. 182, pp. 107-121, 2022. [26]F. Tuz Zohra Anny and O. Islam, “Sentiment polarity classification: A comprehensive approach,” International Journal of Computer Applications, vol. 184, no. 25, pp. 15-21, 2022. [27]F. Tuz Zohra Anny and O. Islam, “Sentiment analysis at sentence and review level,” IEEE Transactions on Computational Social Systems, vol. 9, no. 3, pp. 765-778, 2022. [28]B. Lin, N. Cassee, A. Serebrenik, G. Bavota, N. Novielli, and M. Lanza, “Opinion mining for software engineering: A systematic literature review,” ACM Transactions on Software Engineering and Methodology, vol. 31, no. 3, pp. 1-58, 2022. [29]V. Umarani, A. Julian, and J. Deepa, “Sentiment analysis using various machine learning and deep learning techniques,” Journal of the Nigerian Society of Physical Sciences, vol. 3, no. 4, pp. 385-394, 2021. [30]M. Kabir, M. M. J. Kabir, S. Xu, and B. Badhon, “Sentiment analysis using machine learning approaches: A comprehensive study,” International Journal of Computer Applications, vol. 182, no. 14, pp. 15-19, 2021. H. He, G. Zhou, and S. Zhao, “Exploring e-commerce product experience based on fusion sentiment analysis method,” IEEE Access, vol. 10, pp. 110248-110260, 2022.

© 2025 International Journal of Engineering and Techniques (IJET).

Digital Object Identifier (DOI)

IJET assigns a unique DOI to every accepted article via Zenodo for persistent linking and citation. Your article’s DOI: https://doi.org/{{doi}}

Open DOI About Zenodo DOI

Close