Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{198167,
        author = {Bhargav Sawant and Muiz Tanki and Omkar Pawar and Sunil Patil},
        title = {Explainable AI For Sentiment Analysis On Movie Reviews},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {14734-14741},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=198167},
        abstract = {Sentiment analysis is widely used to interpret public opinion in sectors like shopping, social media and finance. The old ways of doing it have trouble with things like sarcasm, irony and figuring out what words mean in different contexts, which means they do not always give the right answers when looking at real data. This paper is about a way of doing sentiment analysis that combines a lot of different models. 50 Flatten-based networks, 50 LSTM models, 65 GRU models, 5 BERT transformers and 5 RoBERTa transformers. To get a better understanding of what people mean when they write movie reviews. Each and every model was trained separately on a set of movie reviews from IMDB, which has 40,000 labelled reviews. The text was cleaned up using preprocessing methods. Prepared for the models to use. Then the predictions from all models were combined using a stacking ensemble technique, and an XGBoost Stacking was the best at making accurate predictions, with an accuracy of 89.86%. To test and explain the model's working, we used SHAP and LIME methods, which help us understand what is important in the text and how each prediction was made. The results show that our new way of combining models is better and more reliable than old ways and works better in different situations. This work provides a base for doing sentiment analysis in multiple languages and for using it in real-time applications.},
        keywords = {BERT, Ensemble Learning, Explainable AI, IMDB Dataset, LSTM-GRU, RoBERTa, Sentiment Analysis, SHAP},
        month = {April},
        }