Transfer topic modeling with ease and scalability

The increasing volume of short texts generated on so-cial media sites, such as Twitter or Facebook, creates a great demand for effective and efficient topic mod-eling approaches. While latent Dirichlet allocation (LDA) can be applied, it is not optimal due to its weakness in handling short texts with fast-changing topics and scalability concerns. In this paper, we pro-pose a transfer learning approach that utilizes abun-dant labeled documents from other domains (such as Yahoo! News or Wikipedia) to improve topic mod-eling, with better model fitting and result interpre-tation. Specifically, we develop Transfer Hierarchical LDA (thLDA) model, which incorporates the label information from other domains via informative pri-ors. In addition, we develop a parallel implementa-tion of our model for large-scale applications. We demonstrate the effectiveness of our thLDA model on both a microblogging dataset and standard text collections including AP and RCV1 datasets.