Sentimental Analysis For Product Reviews Using Self-Attention Neural Networks

Abstract

Sentiment analysis is now a vital technique for figuring out what customers think about products. The use of self-attention neural networks to analyze the sentiment of product reviews is investigated in this thesis. Self-Attention Neural Networks, in contrast to conventional models, offer a way to assess the relative weight of each word in a phrase, better reflecting the subtleties of natural language. This paper chronicles the development of sentiment analysis approaches and emphasizes the role of self-attention in natural language processing through an extensive survey of the literature. The preprocessing procedures, the architecture of the suggested Self-Attention Neural Network model, and the data collecting from online review sites are all covered in detail in the methods section. The training procedure, hyperparameter adjustment, and assessment metrics that are used to gauge model performance are all covered in the experimental phase. The findings show that Self-Attention Neural Networks execute more accurately and efficiently than traditional models, offering deeper insights into customer sentiment. The conclusion highlights how revolutionary Self-Attention Neural Networks can be for sentiment analysis and makes recommendations for further research.