Topics over time: a non-Markov continuous-time model of topical trends Export

@inproceedings{citeulike:1459200, abstract = {This paper presents an LDA-style topic model that captures not only the low-dimensional structure of data, but also how the structure changes over time. Unlike other recent work that relies on Markov assumptions or discretization of time, here each topic is associated with a continuous distribution over timestamps, and for each generated document, the mixture distribution over topics is influenced by both word co-occurrences and the document's timestamp. Thus, the meaning of a particular topic can be relied upon as constant, but the topics' occurrence and correlations change significantly over time. We present results on nine months of personal email, 17 years of NIPS research papers and over 200 years of presidential state-of-the-union addresses, showing improved topics, better timestamp prediction, and interpretable trends.}, address = {New York, NY, USA}, author = {Wang, Xuerui and McCallum, Andrew}, booktitle = {KDD '06: Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining}, citeulike-article-id = {1459200}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1150450}, citeulike-linkout-1 = {http://dblp.uni-trier.de/rec/bibtex/conf/kdd/WangM06}, citeulike-linkout-2 = {http://dx.doi.org/10.1145/1150402.1150450}, doi = {10.1145/1150402.1150450}, isbn = {1-59593-339-5}, keywords = {topicsovertime}, location = {Philadelphia, PA, USA}, pages = {424--433}, posted-at = {2007-07-16 14:00:27}, priority = {0}, publisher = {ACM}, title = {Topics over time: a non-Markov continuous-time model of topical trends}, url = {http://dx.doi.org/10.1145/1150402.1150450}, year = {2006} }

TY - CONF ID - citeulike:1459200 L3 - citeulike-article-id:1459200 N2 - This paper presents an LDA-style topic model that captures not only the low-dimensional structure of data, but also how the structure changes over time. Unlike other recent work that relies on Markov assumptions or discretization of time, here each topic is associated with a continuous distribution over timestamps, and for each generated document, the mixture distribution over topics is influenced by both word co-occurrences and the document's timestamp. Thus, the meaning of a particular topic can be relied upon as constant, but the topics' occurrence and correlations change significantly over time. We present results on nine months of personal email, 17 years of NIPS research papers and over 200 years of presidential state-of-the-union addresses, showing improved topics, better timestamp prediction, and interpretable trends. CY - Philadelphia, PA, USA SN - 1-59593-339-5 AD - New York, NY, USA EP - 433 TI - Topics over time: a non-Markov continuous-time model of topical trends PB - ACM T2 - KDD '06: Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining SP - 424 KW - topicsovertime AU - Wang, Xuerui AU - McCallum, Andrew PY - 2006/// UR - http://dx.doi.org/10.1145/1150402.1150450 ER -