Customer experience discovery model based on sentiment analysis and machine learning method

Abstract

Exploiting customer insights plays a key role in the long-term development strategy of every business, which strongly supports them in building trust with customers and enhancing competitive advantages. Especially in the period of rapid growth of technology, as there is a significant increase in online users, all the interactions and comments become more important in extracting valuable insights into customer behaviours and emotions. Our research aims to conduct sentiment analysis on comments in mobile commerce apps , then classify them into positive or negative sentiment, based on a hybrid approach that combines supervised machine learning methods and natural language processing techniques; at the same time, evaluating the model performance through the Confusion Matrix to ensure the reliability of the results. The study experimented on a proposed model with 04 machine learning methods including Naive Bayes, SVM, Random Forest and Logistic Regression on more than 935,000 comments collected from 04 popular mobile commerce apps in Vietnam (Tiki, Shopee, Lazada and Sendo). The experiment categorised positive and negative views with high accuracies of 91%, expressed by reports and charts that reflect customer trends and feelings. Moreover, the study also brings new and deeper perspectives on customer behaviours, assisting administrators to detect the strengths and weaknesses of services and apps, thereby improving user experience. Based on the research results, E-commerce businesses can analyse market trends and explore customers' needs and interests to develop effective product and service development strategies.