<p>Aspect Based Sentiment analysis is critical in determining learner sentiments and gaining insights into user attitudes toward online learning platforms. This study proposes a two-stage framework for Aspect-Oriented Sentiment Prediction in Online Education using Explainable AI, specifically the LIME (Local Interpretable Model-Agnostic Explanations) technique. In first stage, a Support Vector Machine classification model is employed to categorize student reviews into + ve, −ve, or neu sentiments. The model reached to a classification accuracy of 0.92, although the performance matrix indicates class imbalance even after applying balancing techniques. In the second stage, LIME is used to interpret the Support Vector Machine model’s predictions, addressing the challenge of understanding black-box model decisions. To ensure local fidelity and interpretability, two core equations are utilized: the Loss Function enforces local accuracy by minimizing prediction discrepancies near the target instance, and the Explanation Model Equation balances simplicity and faithfulness to the original model. A surrogate interpretable model is trained to mimic the SVM’s local behaviour, identifying the input features that most influenced a given sentiment prediction. LIME effectively highlights aspect-specific content that significantly impacts the sentiment classification. For example, in reviews classified as positive, terms such as “good,” “excellent,” and “great” were found to have the highest contribution, particularly in the content-related aspect of online education. This approach not only improves sentiment classification but also provides transparent, actionable insights for educational platforms.</p>