User feedback, expressed through textual reviews and numerical star ratings, is critical in improving mobile applications used in e-governance. However, a common challenge arises when the sentiments conveyed in textual reviews do not align with the corresponding star ratings, creating inconsistencies in user feedback. The mismatch compromises the reliability of ratings, making it difficult for developers to accurately assess and enhance application quality. To address the issue, we propose a novel neural network-based approach that predicts star ratings consistent with the sentiments expressed in user reviews. Our method leverages sentiment analysis to extract sentiment percentages from textual feedback and processes them through a neural network to generate more reliable overall ratings. The proposed framework, trained on a dataset of reviews from various government-related applications on the Google Play Store, demonstrates significant improvements in bridging the gap between user sentiments and star ratings. Experimental results show that our approach reduces the discrepancy between predicted ratings and actual user sentiments, providing developers and policymakers with a dependable tool for interpreting user feedback. The accuracy of feedback analysis improves significantly with this advancement, providing actionable insights for enhancing the quality and user satisfaction of e-governance applications.