Sentiment Analysis of E-commerce Customer Reviews Based on Natural Language Processing
Xiaoxin Lin
Abstract
E-commerce can largely boost the economic development and customer behavior analysis is necessary for e-commerce marketing strategy. We used the dataset of Women's E-Commerce Clothing Reviews to study the sentiment analysis of customer recommendation. Five popular machine learning algorithms were applied to solve the problem, including Logistic Regression, Support Vector Machine (SVM), Random Forest, XGBoost and LightGBM. These algorithms aimed at figuring out the insight correlation between review features and product recommendation based on natural language processing (NLP). The best result was achieved by LightGBM algorithm with highest AUC value and accuracy. The precision, recall and F1 score were all 0.97. Ridge Regression, Linear Kernel SVM and XGboost algorithms which had close performances with the accuracy of 0.94. This research can help generate a deeper comprehension of customer sentiment and grasp customer psychology in e-commerce transaction industry