Optimized Gaussian Deep Learning Architecture With Sentimental Analysis For The E-Commerce Application

Abstract

In the context of E-commerce, sentiment analysis plays a crucial role in understanding and evaluating the subjective information present in customer reviews, feedback, and comments.  With implementation of sentimental analysis automatically identify and classify sentiments as positive, negative, or neutral, providing valuable insights into how customers perceive products, services, or overall experiences. Sentiment analysis in E-commerce encounters challenges such as contextual ambiguity, difficulty in detecting sarcasm, data imbalance with an abundance of positive reviews, domain-specific language nuances, multilingual diversity, adapting to various product types, accounting for temporal shifts in sentiments, addressing privacy concerns, tackling review spam, and ensuring effective integration into business decisions. This paper proposed a novel approach to sentiment analysis in E-commerce utilizing the Gaussian Long Short-Term Memory (G-LSTM) model. The proposed G-LSTM model implements Fejer Kernel filter for the data pre-processing followed by the dictionary-based model for the feature extraction process. The feature selection is performed with the optimization of the features such as Seahorse Optimization mode. Finally, the classification is performed with the Gaussian Long Short Term Memory (LSTM) model for the classification. The G-LSTM model showcases remarkable performance, achieving an accuracy of 98.3% along with superior precision, recall, and F1-Score compared to baseline LSTM, SVM, and Random Forest models.