A topic model based on poisson decomposition

Abstract

Determining appropriate statistical distributions for modeling text corpora is important for accurate estimation of numerical charac- teristics. Based on the validity of the test on a claim that the data conforms to Poisson distribution we propose Poisson decomposi- tion model (PDM), a statistical model for modeling count data of text corpora, which can straightly capture each document’s mul- tidimensional numerical characteristics on topics. In PDM, each topic is represented as a parameter vector with multidimensional Poisson distribution, which can be easily normalized to multino- mial term probabilities and each document is represented as mea- surements on topics and thereby reduced to a measurement vec- tor on topics. We use gradient descent methods and sampling al- gorithm for parameter estimation. We carry out extensive experi- ments on the topics produced by our models. The results demon- strate our approach can extract more coherent topics and is com- petitive in document clustering by using the PDM-based features, compared to PLSI and LDA.