Sentiment Analysis for e-Payment Service Providers Using Evolutionary eXtreme Gradient Boosting

Abstract

Online services depend primarily on customer feedback and communications. When this kind of input is lacking, the overall approach of the service provider can shift in unintended ways. These services rely on feedback to maintain consumer satisfaction. Online social networks are a rich source of consumer data related to services and products. Well developed methods like sentiment analysis can offer insightful analyses and aid service providers in predicting outcomes based on their reviews-which, in turn, enables decision-makers to develop effective strategic plans. However, gathering this data is more challenging on Arabic online social networks, due to the complexity of the Arabic language and its dialects. In this study, we propose an approach to sentiment analysis that combines a neutrality detector model with eXtreme Gradient Boosting and a genetic algorithm to effectively predict and analyze customers' opinions of an e-Payment service through an Arabic social network. The proposed approach yields excellent results compared to other approaches. Feature analysis is also conducted on consumer reviews to identify influencing keywords.

Keywords

• Evolutionary
• genetic algorithm
• neutrality detector model
• sentiment analysis
• social network
• XGBoost