Rating Prediction Based on TripAdvisor Reviews

Rating is becoming increasingly important for customers, which helps them obtain useful information of products more easily and more efficiently. Normally, a customer rates products from different aspects and then makes an overall rating. According to this human thinking process, we are curious about how ratings in different aspects would influence the final overall ratings. Therefore, we make two different prediction models: 1. Use review context to predict overall rates predictions. 2. Use review context to predict rates in different aspects(cleanliness, location, rooms, service, sleep quality and value) and then use these ratings to predict the overall ratings. By comparing MSE of two models, we found out that the first model is more accurate than the second one. Keywords—linear model, rating predictions