Deep Reinforcement Learning for Ethically‐Aware Personal‎ized Sentiment Analysis of Customer Reviews

  • Authors

    • Dr. N. Kannaiya Raja Post Doctoral Researcher, Lincoln University College, Malaysia
    • Dr. Pawan Kumar Chaurasia Associate Professor, Department of Information Technology, Babasaheb Bhimrao Ambedkar Central University, ‎Lucknow, Uttar Pradesh, India ‎
    • Prof Dr. Midhunchakkaravarthy Professor, Lincoln University College, Malaysia
    https://doi.org/10.14419/awm5r821

    Received date: June 27, 2025

    Accepted date: August 3, 2025

    Published date: August 14, 2025

  • Deep Reinforcement Learning; Ethically-Aware Sentiment Analysis; Personalized Sentiment Analysis; Aspect-Based Sentiment Analysis; ‎GRU Policy Net-works‎.

  • Abstract

    A conventional sentiment analyzer is static, labeled corpora and risk amplifying biases entrenched in its training data. To overcome these ‎limitations, we introduce a multi‐agent, aspect‐based deep reinforcement learning framework, the ADRSA algorithm, which continuously ‎adapts to streaming customer reviews while enforcing ethical personalization. Each agent specializes in one product aspect, for example, price, ‎food, product, and updates its policy network (GRU + policy head) in real time, receiving composite rewards for both classification accuracy ‎and adherence to fairness constraints. We evaluated ADRSA on 50K real‐world reviews (Amazon Electronics, Books, Home & Kitchen) ‎with anonymized age, gender, and region metadata. On the “price” aspect (APA), ADRSA attains 0.75 precision, 0.72 recall, 0.78 F1 and ‎‎0.88 accuracy; on “food” (AFR), 0.74 precision, 0.60 recall, 0.65 F1 and 0.89 accuracy, and on “product” (APR), 0.75 precision, 0.62 re-‎call, 0.65 F1 and an exceptional 0.98 accuracy, outperforming static LSTM (max 0.91 acc) and RNMS and SVM baselines by up to 7 per-‎centage points in F1. These results demonstrate ADRSA’s ability to deliver fine‐grained, responsible sentiment insights that evolve with ‎new user language while safeguarding fairness across demographics‎.

  • References

    1. Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, May 2015. https://doi.org/10.1038/nature14539.
    2. V. Mnih et al., “Human-level control through deep reinforcement learning,” Nature, vol. 518, no. 7540, pp. 529–533, Feb. 2015. https://doi.org/10.1038/nature14236.
    3. D. Silver et al., “Mastering the game of Go with deep neural networks and tree search,” Nature, vol. 529, no. 7587, pp. 484–489, Jan. 2016. https://doi.org/10.1038/nature16961.
    4. Z. Wang, T. Schaul, M. Hessel, H. Van Hasselt, M. Lanctot, and N. de Freitas, “Dueling network architectures for deep reinforcement learning,” in Proc. 33rd Int. Conf. Machine Learning, New York, NY, USA, 2016, pp. 1995–2003.
    5. T. Takanobu, P. Liu, and K. Huang, “Hierarchical reinforcement learning for joint entity and relation extraction,” in Proc. AAAI Conf. Artificial In-telligence, vol. 33, 2019, pp. 6370–6377. https://doi.org/10.1609/aaai.v33i01.33017072.
    6. Y. Feng, P. Quan, J. Huang, and X. Xue, “Reinforcement learning for relation classification from noisy data,” in Proc. ACL, 2018, pp. 1627–1636. https://doi.org/10.18653/v1/P18-1157.
    7. Z. Qin, Y. He, and J. Huang, “A hierarchical reinforcement learning framework for relation extraction,” in Proc. AAAI, 2019, pp. 5801–5808.
    8. S. Ma, Y. Peng, and J. Cambria, “Targeted aspect‐based sentiment analysis via embedding commonsense knowledge into an attentive LSTM,” in Proc. AAAI, 2018, pp. 5876–5883. https://doi.org/10.1609/aaai.v32i1.11903.
    9. C. Aydın and S. Güngör, “Aspect‐based sentiment analysis using recursive and recurrent neural networks,” Expert Systems with Applications, vol. 150, 2020, Art. no. 113287. https://doi.org/10.1016/j.eswa.2020.113287.
    10. X. Qiu, Y. Qian, Z. Han, and Y. Li, “Deep Q‐network hashing for sequential data retrieval,” IEEE Trans. Multimedia, vol. 22, no. 11, pp. 2865–2877, Nov. 2020.
    11. S. Krening, R. K. Singh, and S. Thrun, “An agent‐centric sentiment filter for improving video game AI,” in Proc. IROS, 2017, pp. 1605–1611. https://doi.org/10.1109/IROS.2017.8205953.
    12. T. Chu, J. Tan, and P. Zhao, “Actor‐centric A2C for adaptive traffic signal control,” IEEE Trans. Intelligent Transportation Systems, early access, 2020.
    13. R. Pröllochs, R. Feuerriegel, and R. Neumann, “Automated negation detection in sentence polarity classification,” Information Systems, vol. 92, 2020, Art. no. 101513.
    14. K. Bekoulis, E. Augenstein, D. Vlachos, and S. Cristea, “Joint entity and relation extraction with a hybrid neural network,” in Proc. EMNLP, 2018, pp. 1014–1024. https://doi.org/10.18653/v1/D18-1126.
    15. C. Nguyen et al., “Clause‐level structure‐aware sentiment analysis with reinforcement learning,” Information Processing & Management, vol. 57, no. 5, 2020, Art. no. 102288. https://doi.org/10.1016/j.ipm.2020.102288.
    16. X. Mao, S. K. Sethi, W. Xu, and L. Jin, “Attention‐based RNN for aspect‐level sentiment analysis,” in Proc. ACL, 2024, pp. 221–231.
    17. R. Kharde and S. Sonawane, “Sentiment analysis of Twitter data: A survey of techniques,” International Journal of Computer Applications, vol. 47, no. 17, pp. 25–34, Jun. 2016. https://doi.org/10.5120/ijca2016910672.
    18. H. Kuang, J. Wang, and X. Liu, “Opinion mining of consumer reviews on Twitter: A comparative study,” Expert Systems with Applications, vol. 205, 2023, Art. no. 117407.
    19. M. Wadden et al., “Entity, relation, and event extraction with contextualized span representations,” in Proc. EMNLP/IJCNLP, 2019, pp. 5783–5788. https://doi.org/10.18653/v1/D19-1588.
    20. T. Ren, Y. Zhou, Q. Jiang, and J. Tang, “Deep curricular reinforcement learning for adaptive curriculum sequencing,” in Proc. AAAI, 2018, pp. 4902–4909. https://doi.org/10.1609/aaai.v32i1.12203.
    21. D. Feng, X. Jin, Y. Jin, and Q. Wang, “Abstractive review summarization with topic modeling and reinforcement learning,” in Proc. COLING, 2020, pp. 122–132. https://doi.org/10.18653/v1/2020.coling-main.11.
    22. Jiang, Y., & Nachum, O. (2019). Identifying and correcting label bias in machine learning. arXiv preprint arXiv:1901.04966. https://arxiv.org/abs/1901.04966.
    23. Jabbari, S., Joseph, M., Kearns, M., Morgenstern, J., & Roth, A. (2017). Fairness in Reinforcement Learning. In Proceedings of the 34th Interna-tional Conference on Machine Learning (ICML), PMLR 70:1617-1626. https://proceedings.mlr.press/v70/jabbari17a.html.
    24. Wu, X., Wang, Z., Li, J., & Li, H. (2021). Fairness-aware reinforcement learning: A survey. IEEE Transactions on Knowledge and Data Engineer-ing, 34(9), 4403-4419.
    25. Miyato, T., Dai, A. M., & Goodfellow, I. (2017). Adversarial training methods for semi-supervised text classification. International Conference on Learning Representations (ICLR).
    26. Zhao, J., Wang, T., Yatskar, M., Ordonez, V., & Chang, K. W. (2017). Men also like shopping: Reducing gender bias amplification using corpus-level constraints. Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2979–2989. https://doi.org/10.18653/v1/D17-1323.
    27. Oyelade, O. N., & Oladipupo, O. O. (2013). Application of data mining techniques in customer relationship management using social networks. International Journal of Computer Applications, 62(3), 15–19.

  • Downloads

  • How to Cite

    Raja, D. N. K. . ., Chaurasia, D. P. K. ., & Midhunchakkaravarthy , P. D. . (2025). Deep Reinforcement Learning for Ethically‐Aware Personal‎ized Sentiment Analysis of Customer Reviews. International Journal of Basic and Applied Sciences, 14(4), 430-438. https://doi.org/10.14419/awm5r821