Mining Online Reviews with Fake Review Detection to Improve Logistics Service Quality

This study proposes an analytical framework to identify important information in online reviews and uses the service quality models and relevant literature on logistics service quality to help construct and analyze the logistics service quality topics. First, this study establishes a fake review detection model based on the convolution neural network (CNN) model to remove potential fake reviews. Then, the SERVQUAL and E-S-QUAL models are used to assist in constructing a Latent Dirichlet Allocation (LDA) topic model to identify relevant topics mentioned in the online reviews about logistics service quality. In further data analysis, this study segments the comments, calculates the sentiment scores of the sentences that mention the logistics service quality in the online reviews, and uses the number of helpfulness votes of each online review to calculate the weight of the sentiment score.