EPR-ML: E-Commerce Product Recommendation Using NLP and Machine Learning Algorithm

Abstract

Using tags and other forms of textual information, online retailers can create their product listings, descriptions, and categories. E-commerce information services, such as search and product recommendation, depend significantly on textual features to assist buyers in finding the items they want. This research focuses on “tags,” which often use textual descriptions of items. We assume that merchants are not always the “best” suppliers of item tag information, either because they are ill-equipped to do so (since they have not been “trained”) or because they are purposefully attempting to rig the system by using misleading or erroneous tags to sell their commodities (tag spam). To address these concerns, we may use automated tag recommendation techniques to enhance the precision with which we suggest tags for every specific product. We proposed EPR-ML for E-commerce product recommendation using NLP and ML algorithms. This research employed a product sentiment dataset normalized using NLP; the best features were selected using Logistic regression (LR). The classification was performed using various machine learning algorithms, including Linear support vector machine (L- SVM) and Gaussian nave Bayes (GNB), to determine which model is most accurate at predicting the number of days it will take a video to trend from the time it was uploaded and the number of days it will trend on the trending list. Using LSVM, the research achieved a maximum accuracy of 96%.