A comprehensive E-commerce customer behavior analysis using convolutional methods. (December 2021)

• Record Type: Journal Article
• Authors: Nagrath, Preeti; Nguyen, Tu N.; Aggarwal, Shivani; Hemanth, D Jude
• Is Part Of: Computers & electrical engineering. Volume 96:Part B(2021)
• Publication Date: 2021-12
• Subjects: Sentiment analysis -- Capsule network -- Attention -- Recurrent neural networks -- Convolutional neural networks -- Long short-term memory -- Gated recurrent units; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2021.107541
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: E-commerce is one of the most widely preferred business models all around the world. People can order items and have them in days or maybe in hours without visiting the shops. They count on the reviews of the online items to purchase them, so reviews have become one of the most significant metrics. Modern day algorithms and models can be made to decide the reviews seen by the user to judge an item fairly. We have attempted to make models that can judge the sentiment of the user writing the reviews from the text of the review, classifying it into positive, negative or neutral. Edge Computation is an emerging field in this scenario that can be very beneficial alongside capsule networks for processing the data as data capsules. In this work, we have performed a comparison of the effect of convolutional layers, attention layers and capsule network layers on base models for GRU, LSTM, Bi-GRU, and Bi-LSTM. These are done with the purpose to support the functioning of Edge Computation. The final results show that Bi-LSTM-Capsule Model (BiCapsModel) gives highest accuracy of 94.8%. The final outcome is very promising and can be used in various domains in journalism, product review and stock market. For classification, an Amazon review dataset has been considered. The dataset has various types of classified reviews based on which the emotional text analysis can be performed efficiently covering each aspect.