LPTAAProbabilisticModelfor.PDFVIP

LPTA: A Probabilistic Model for Latent Periodic Topic Analysis Zhijun Yin1, Liangliang Cao2, Jiawei Han1, Chengxiang Zhai1, Thomas Huang3 1Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL, USA 2IBM T. J. Watson Research Center, Hawthorne, NY, USA 3Department of ECE and Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL, USA zyin3@, liangliang.cao@, hanj@, czhai@, t-huang1@ Abstract—This paper studies the problem of latent periodic topic analysis from timestamped documents. The examples of timestamped documents include news articles, sales records, financial reports, TV programs, and more recently, posts from social media websites such as Flickr, Twitter, and Facebook. Different from detecting periodic patterns in traditional time series database, we discover the topics of coherent semantics and periodic characteristics where a topic is represented by a distribution of words. We propose a model called LPTA (Latent Periodic Topic Analysis) that exploits the periodicity of the terms as well as term co-occurrences. To show the effectiveness of our model, we collect several representative datasets including Seminar, DBLP and Flickr. The results show that our model can discover the latent periodic topics effectively and leverage the information from both text and time well. Keywords-periodic topics; topic modeling; I. INTRODUCTION Periodic phenomena exist ubiquitously in our lives, and lots of natural and social topics have periodic recurring patterns. Hurricanes strike over the similar seasons ev- ery year. Many music and film festivals are held during similar periods annually. Sales offered by different brands culminate during Thanksgiving and Christmas every year. TV programs usually follow weekly schedules. Publicly traded companies are required to disclose information on an ongoing basis by submitting both annual reports and quarterly reports. Due to the prevalent existence of periodic topics, periodicity a