Sentiment Analysis for E-Commerce Product Reviews Using CNN-LSTM

Due to the exponential growth of the number of user-generated reviews on e-commerce platforms, sentiment analysis has emerged for analyzing consumer opinions. This study aims to classify consumer sentiment by analyzing reviews from a product review dataset collected from a public repository. The aim is to develop a novel hybrid convolutional neural network (CNN) - long short-term memory (LSTM) model for classifying product reviews into positive, negative, and neutral sentiment categories. The proposed methodology of this study includes data preprocessing and feature extraction via natural language processing techniques to ensure that the data that are fed into the model can capture intricate patterns for the classification of sentiments. The CNN-LSTM hybrid model was employed for the sentiment classification of customer reviews because it combines the architecture of CNNs to handle spatial dependencies and LSTMs to address long-term dependencies, which leads to increased overall performance and efficiency in the text data. The findings indicate that the hybrid CNN-LSTM model significantly outperforms traditional methods and other state-of-the-art architectures in terms of accuracy, precision, recall, and an F1 score of 98%. This investigation contributes to the field by providing evidence from empirical studies on an effective approach toward sentiment classification, which can further assist business organizations in making strategic decisions on the basis of customer feedback.