The Use of Topic Modeling in Mining Customers’ Reviews

Abstract

Social media is becoming one of the most influential tools in this century. Users of social media platforms generate a vast amount of content that can be used by the end user to analyze, measure, monitor, and interpret people's thoughts, beliefs, opinions, feelings, and even relationships. Given the huge impact of social media on business performance, companies are creating their own platforms to have more channels through which they can gain insights into customers' views on products and services. Social media is also becoming an important channel in restaurateurs' communication with current and potential customers. A customer can leave a comment to share their satisfaction or dissatisfaction and rate the overall experience with the product or service received. In this study, Latent Dirichlet Allocation (LDA) modeling algorithm was used to extract the main themes in the customer feedbacks. The extracted themes revealed that food quality, service quality, speed and completeness of order are the major themes raised by the customers. Analyzing the e-WOM of the customers enables the management to gain insights into the business process, product and service. Such analysis can help provide better personalized service to improve business performance. Keywords— online reviews; text mining; Topic Modeling; Latent Dirichlet Allocation (LDA); Social Media; sentiment analysis.