Detecting Topic Drift with Compound Topic Models

Dan Knights; Michael C. Mozer; Nicolas Nicolov

doi:10.1609/icwsm.v3i1.13982

Detecting Topic Drift with Compound Topic Models

Dan Knights
, Michael C. Mozer
, Nicolas Nicolov

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

18 Scopus citations

Abstract

The Latent Dirichlet Allocation topic model of Blei, Ng, & Jordan (2003) is well-established as an effective approach to recovering meaningful topics of conversation from a set of documents. However, a useful analysis of user-generated content is concerned not only with the recovery of topics from a static data set, but with the evolution of topics over time. We employ a compound topic model (CTM) to track topics across two distinct data sets (i.e. past and present) and to visualize trends in topics over time; we evaluate several metrics for detecting a change in the distribution of topics within a time-window; and we illustrate how our approach discovers emerging conversation topics related to current events in real data sets.

Original language	English (US)
Title of host publication	Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009
Publisher	AAAI press
Pages	242-245
Number of pages	4
Edition	1
ISBN (Electronic)	9781577354215
DOIs	https://doi.org/10.1609/icwsm.v3i1.13982
State	Published - May 20 2009
Externally published	Yes
Event	3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009 - San Jose, United States Duration: May 17 2009 → May 20 2009

Publication series

Name	Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009
Number	1
Volume	3

Conference

Conference	3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009
Country/Territory	United States
City	San Jose
Period	5/17/09 → 5/20/09

Bibliographical note

Publisher Copyright:
Copyright © 2009, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.

Publisher link

10.1609/icwsm.v3i1.13982

Find It

Find It at U of M Libraries

Cite this

Knights, D., Mozer, M. C., & Nicolov, N. (2009). Detecting Topic Drift with Compound Topic Models. In Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009 (1 ed., pp. 242-245). (Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009; Vol. 3, No. 1). AAAI press. https://doi.org/10.1609/icwsm.v3i1.13982

Detecting Topic Drift with Compound Topic Models. / Knights, Dan; Mozer, Michael C.; Nicolov, Nicolas.
Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009. 1. ed. AAAI press, 2009. p. 242-245 (Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009; Vol. 3, No. 1).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Knights, D, Mozer, MC & Nicolov, N 2009, Detecting Topic Drift with Compound Topic Models. in Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009. 1 edn, Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009, no. 1, vol. 3, AAAI press, pp. 242-245, 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009, San Jose, United States, 5/17/09. https://doi.org/10.1609/icwsm.v3i1.13982

@inproceedings{786eca560c104bdd8782bde626395d9a,

title = "Detecting Topic Drift with Compound Topic Models",

abstract = "The Latent Dirichlet Allocation topic model of Blei, Ng, \& Jordan (2003) is well-established as an effective approach to recovering meaningful topics of conversation from a set of documents. However, a useful analysis of user-generated content is concerned not only with the recovery of topics from a static data set, but with the evolution of topics over time. We employ a compound topic model (CTM) to track topics across two distinct data sets (i.e. past and present) and to visualize trends in topics over time; we evaluate several metrics for detecting a change in the distribution of topics within a time-window; and we illustrate how our approach discovers emerging conversation topics related to current events in real data sets.",

author = "Dan Knights and Mozer, \{Michael C.\} and Nicolas Nicolov",

note = "Publisher Copyright: Copyright {\textcopyright} 2009, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.; 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009 ; Conference date: 17-05-2009 Through 20-05-2009",

year = "2009",

month = may,

day = "20",

doi = "10.1609/icwsm.v3i1.13982",

language = "English (US)",

series = "Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009",

publisher = "AAAI press",

number = "1",

pages = "242--245",

booktitle = "Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009",

edition = "1",

}

TY - GEN

T1 - Detecting Topic Drift with Compound Topic Models

AU - Knights, Dan

AU - Mozer, Michael C.

AU - Nicolov, Nicolas

PY - 2009/5/20

Y1 - 2009/5/20

N2 - The Latent Dirichlet Allocation topic model of Blei, Ng, & Jordan (2003) is well-established as an effective approach to recovering meaningful topics of conversation from a set of documents. However, a useful analysis of user-generated content is concerned not only with the recovery of topics from a static data set, but with the evolution of topics over time. We employ a compound topic model (CTM) to track topics across two distinct data sets (i.e. past and present) and to visualize trends in topics over time; we evaluate several metrics for detecting a change in the distribution of topics within a time-window; and we illustrate how our approach discovers emerging conversation topics related to current events in real data sets.

AB - The Latent Dirichlet Allocation topic model of Blei, Ng, & Jordan (2003) is well-established as an effective approach to recovering meaningful topics of conversation from a set of documents. However, a useful analysis of user-generated content is concerned not only with the recovery of topics from a static data set, but with the evolution of topics over time. We employ a compound topic model (CTM) to track topics across two distinct data sets (i.e. past and present) and to visualize trends in topics over time; we evaluate several metrics for detecting a change in the distribution of topics within a time-window; and we illustrate how our approach discovers emerging conversation topics related to current events in real data sets.

UR - https://www.scopus.com/pages/publications/84900675005

UR - https://www.scopus.com/pages/publications/84900675005#tab=citedBy

U2 - 10.1609/icwsm.v3i1.13982

DO - 10.1609/icwsm.v3i1.13982

M3 - Conference contribution

AN - SCOPUS:84900675005

T3 - Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009

SP - 242

EP - 245

BT - Proceedings of the 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009

PB - AAAI press

T2 - 3rd International AAAI Conference on Weblogs and Social Media, ICWSM 2009

Y2 - 17 May 2009 through 20 May 2009

ER -

Detecting Topic Drift with Compound Topic Models

Abstract

Publication series

Conference

Bibliographical note

Publisher link

Find It

Other files and links

Fingerprint

Cite this