dynamic topic models - research showcase @ cmu.pdfVIP

Carnegie Mellon University Research Showcase @ CMU Computer Science Department School of Computer Science 6-2006 Dynamic Topic Models David M. Blei Princeton University J ohn D. Lafferty Carneg ie Mellon University Follow th s and additional works at : /compsc Publ shed In Proceedings of the 23rd international Conference on Machine Learning. ICML 06, 113-120. Th s Conference Proceeding s brought to you for free and open access by the School of Computer Science at Research Showcase @ CMU. It has been accepted for inclusion in Computer Science Department by an author zed admin strator of Research Showcase @ CMU. For more information, please contact research-showcase@. Dynamic Topic Models David M. Blei BLEI @CS .PRINCETON .EDU Computer Science Department, Princeton University, Princeton, NJ 08544, USA John D. Lafferty LAFFERTY @CS .CMU .EDU School of Computer Science, Carnegie Mellon University, Pittsburgh PA 15213, USA Abstract ment are assumed to be independently drawn from a mix- A family of probabilistic time series models is ture of multinomials. The mixing proportions are randomly developed to analyze the time evolution of topics drawn for each document; the mixture components, or top- in large document collections. The approach is ics, are shared by all documents. Thus, each document to use state space models on the natural param- reflects the components with different proportions. These eters of the multinomial distributions that repre- models are a powerful method of dimensional