Novel Customer Review Analysis System Based On Balanced Deep Review And Rating Differences In User Preference
Khaja Miftah Uddin, Shaik Salman, Mohammad Baseer, Dr.Abdul ahad afroz · International jounal of information technology and computer engineering. · 2025
The swift expansion of online ecommerceplatforms and mobile applications hasmade it simpler to collect vast volumes of data,offering insightful information about customerbehavior. Helping consumers make decisionsabout what to buy now requires analyzing userreviews. In the suggested system, we present asolution by integrating a CNN (ConvolutionalNeural Network) model for reviewcategorization with NLP (Natural LanguageProcessing) approaches. To enhance itscomprehension of the subtleties of reviewcontent, the model integrates word embedding,tokenization, and text preprocessing approaches.The CNN-based architecture greatly increasesprediction accuracy and processing efficiency byenhancing the capacity to identify key patternsand correlations in the data. This methodprovides a more accurate and scalable model forreview analysis, hence overcoming theshortcomings of earlier approaches. It can bereadily modified to manage varied textualmaterial and large-scale datasets. Thesuggested system outperforms the currentmethods in terms of classification, according toexperimental evaluation. The approachincreases decision-making confidence in ecommerceplatforms and predicts usefulevaluations by concentrating on importantpatterns and correlations in the text data.