Predicting Customer Loyalty from E-Commerce Reviews Using Aspect-Based Sentiment Analysis and ANN

Abstract

oday’s generation like to purchase online things. Online market is growing very fast. As more retailers for this market appears, the battle to sell things becomes fiercer. On other hands these marketplaces are establishing the trust of their customers and provide them with handy options. Consumers are much more intelligent; while making a buy, they investigate and evaluate options. Some consumers are still hesitant to make purchases online, while some of them are regular buyers. People becomes very conscious of the necessity of buying online as a result of numerous disadvantages, A system is much needed which provide a proper analysis of regular buyer to facilitate new customer and company. In this research a wise technique is offers to measure customers loyalty to a product it helps news customer to take decision faster and also assist new customers. Our technique employs a unique concept for determining a devotion of a buyer to a particular brand or item, and it may be of assistance to a new customer in making a choice made regarding a certain item based on its many functions and past customer comments. In our proposed model we used artificial neural network (ANN) approach to measure customers loyalty for this purpose a large data set from Kaggle based on customers reviews on online product is taken. POS tagging extract the textual and non-textual information of the reviews, pre-processes them, and converts this textual information into tokens. The proposed ANN approach generates vectors from pre-processed and mapped reviews. For training, the featured dataset is sent into the suggested ANN model. After training another sample data is used to test the suggested approach. For the prediction of loyalty, the trained dataset is used. This research is anticipated to not only contribute to the literature on customer loyalty prediction, but also to provide e- commerce managements on the pursuit of client loyalty.