Mining Hot Topics from Free-Text Customer Reviews An LDA-Based Approach

This study examines how the Latent Dirichlet Allocation (LDA) model combined with natural language processing techniques can be used to identify hot topics from free-text customer reviews. To verify the validity of the proposed approach, 21 580 restaurant reviews are collected. Each review is viewed as a probabilistic mixture of latent topics and each topic is treated as a probability distribution over words in a vocabulary. Parameters are estimated with Gibbs sampling, and the hot topics with top words are acquired. The experiments show that this approach could produce satisfactory results.