DMCA
Topics over time: A non-Markov continuous-time model of topical trends (2006)
Cached
Download Links
- [ciir.cs.umass.edu]
- [maroo.cs.umass.edu]
- [ciir-publications.cs.umass.edu]
- [maroo.cs.umass.edu]
- DBLP
Other Repositories/Bibliography
| Venue: | in SIGKDD |
| Citations: | 251 - 10 self |
BibTeX
@INPROCEEDINGS{Wang06topicsover,
author = {Xuerui Wang and Andrew Mccallum},
title = {Topics over time: A non-Markov continuous-time model of topical trends},
booktitle = {in SIGKDD},
year = {2006}
}
Share
 |
 |
 |
 |
||
OpenURL
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.
Keyphrases
topical trend non-markov continuous-time model low-dimensional structure generated document markov assumption personal email mixture distribution interpretable trend nip research paper word co-occurrence lda-style topic model topic occurrence document timestamp continuous distribution timestamp prediction presidential state-of-the-union address improved topic particular topic present result recent work
