A Context-Aware ELECTRA-Based Dual-Output Framework for Intelligent Mobile Review Title and Rating Prediction

Abstract

The increasing popularity of mobile applications has resulted in a substantial growth in usergenerated reviews and ratings, making them valuable resources for evaluating customer satisfaction and application quality. These reviews contain important information regarding user experiences, feature requests, software performance, and overall product perception. However, manually analysing large volumes of textual feedback is labour-intensive, inconsistent, and incapable of efficiently processing continuously growing datasets. Furthermore, conventional review analysis methods often struggle to capture contextual semantics and handle imbalanced rating distributions, leading to reduced prediction accuracy. To overcome these challenges, this research proposes an intelligent dual-target classification framework for mobile user reviews and ratings. The proposed system initially performs comprehensive Natural Language Processing (NLP) preprocessing and Exploratory Data Analysis (EDA) to improve data quality and examine the characteristics of the review dataset. Subsequently, Efficiently Learning an Encoder that Classifies Token Replacements Accurately (ELECTRA) is employed to generate contextual feature representations that preserve semantic relationships within user reviews. To alleviate class imbalance, the K-Means Synthetic Minority Over-Sampling Technique (K-Means SMOTE) is utilized to generate representative samples for minority classes, thereby improving model generalization. The extracted features are then processed using an Extra Trees Classifier (ETC) to simultaneously predict review titles and corresponding user ratings. Comparative analysis with Adaptive Boosting Classifier (ABC) and Tao Tree Classifier (TTC) demonstrates that the proposed framework achieves superior predictive performance. The developed system provides an efficient and scalable solution for automated mobile review analysis, enabling organizations to understand customer opinions, monitor application quality, and support data-driven product improvement strategies.