NLP-Based Sentiment Analysis And User Satisfaction Evaluation of Solar Energy Monitoring Applications

The increasing adoption of solar energy systems has created a growing demand for digital applications that help users monitor electricity generation, energy consumption, inverter performance and overall system efficiency. However, the usefulness of these applications also depends on user experience and satisfaction. Application-store reviews provide valuable information about user opinions and technical difficulties, but analysing large volumes of unstructured feedback manually is challenging. This study presents an NLP-based framework for analysing user reviews and evaluating satisfaction with solar energy monitoring applications. A dataset containing 1,777 reviews collected from five applications was processed using text preprocessing and Term Frequency–Inverse Document Frequency (TF-IDF) feature extraction. Four machine learning classifiers, namely Multinomial Naive Bayes, Logistic Regression, Support Vector Machine and Random Forest, were evaluated for sentiment classification. Sentiment labels were derived from numerical user ratings, with 1–2 stars classified as negative, 3 stars as neutral and 4–5 stars as positive. The study also applies keyword-based aspect identification to examine eight application features, including interface usability, real-time monitoring, data accuracy, inverter connectivity, notifications, application performance, compatibility and technical support. A provisional AspectLevel Satisfaction Index (ASI) is used to summarize rating-derived feedback associated with these features. Logistic Regression achieved the highest classification accuracy of 75.84%, while Support Vector Machine obtained the highest macro-averaged F1-score of 56.87%. The findings highlight the potential of NLP techniques to identify user concerns and support improvements in solar energy monitoring applications.