Implementation of User Rating Classification for Amazon Food Review Dataset Using SVM and LSTM

Abstract

This research investigates the challenge of classifying user ratings for Amazon food reviews using Support Vector Machines (SVM) and Long Short-Term Memory (LSTM) neural networks. The objective is to forecast the sentiment or user rating categorization of food reviews in order to provide important information for both consumers and vendors on the network. The dataset comprises textual reviews and their related user ratings collected from the Amazon food goods category. A train-test split is conducted in order to train the Support Vector Machine (SVM) model using the training dataset and adjust its hyperparameters to achieve optimal performance. In the context of Long Short-Term Memory (LSTM), the neural network is trained by using the training set and incorporating strategies such as dropout and early stopping to mitigate the issue of overfitting. The empirical findings demonstrate that both Support Vector Machines (SVM) and Long Short-Term Memory (LSTM) models exhibit a notable level of precision when used for the purpose of forecasting user ratings in the context of Amazon food reviews. Support Vector Machines (SVM) have exceptional performance in managing datasets that are both sparse and high-dimensional. On the other hand, Long Short-Term Memory (LSTM) networks are very proficient at capturing contextual connections within textual data. The results provide significant insights for organisations about customer satisfaction and sentiment patterns, enabling them to make informed choices based on data to enhance product offerings and improve customer experiences. In addition, prospective consumers might get advantages from the precise sentiment analysis while evaluating food acquisitions on the Amazon platform.