Online sentiment and topic dynamics tracking over the streaming data

Chenghua Lin, Yulan He, Elizabeth Cano

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

We propose a dynamic joint sentiment-topic model (dJST) which is able to effectively track sentiment and topic dynamics over the streaming data. Both topic and sentiment dynamics are captured by assuming that the current sentiment-topic specific word distributions are generated according to the word distributions at previous epochs. We study three different ways of accounting for such dependency information, (1) Sliding window where the current sentiment-topic-word distributions are dependent on the previous sentiment-topic specific word distributions in the last S epochs; (2) Skip model where history sentiment-topic-word distributions are considered by skipping some epochs in between; and (3) Multiscale model where previous long-and short-timescale distributions are taken into consideration. We derive efficient online inference procedures to sequentially update the model with newly arrived data and show the effectiveness of our proposed model on the Mozilla add-on reviews crawled between 2007 and 2011.
Original languageEnglish
Title of host publicationInternational Conference on Social Computing (SocialCom)
Place of PublicationAmsterdam
PublisherIEEE Computer Society
Pages258-266
Number of pages9
ISBN (Print)978-1-4673-5638-1
DOIs
Publication statusPublished - 2012

Cite this

Lin, C., He, Y., & Cano, E. (2012). Online sentiment and topic dynamics tracking over the streaming data. In International Conference on Social Computing (SocialCom) (pp. 258-266). Amsterdam: IEEE Computer Society. https://doi.org/10.1109/SocialCom-PASSAT.2012.14

Online sentiment and topic dynamics tracking over the streaming data. / Lin, Chenghua; He, Yulan; Cano, Elizabeth.

International Conference on Social Computing (SocialCom). Amsterdam : IEEE Computer Society, 2012. p. 258-266.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Lin, C, He, Y & Cano, E 2012, Online sentiment and topic dynamics tracking over the streaming data. in International Conference on Social Computing (SocialCom). IEEE Computer Society, Amsterdam, pp. 258-266. https://doi.org/10.1109/SocialCom-PASSAT.2012.14
Lin C, He Y, Cano E. Online sentiment and topic dynamics tracking over the streaming data. In International Conference on Social Computing (SocialCom). Amsterdam: IEEE Computer Society. 2012. p. 258-266 https://doi.org/10.1109/SocialCom-PASSAT.2012.14
Lin, Chenghua ; He, Yulan ; Cano, Elizabeth. / Online sentiment and topic dynamics tracking over the streaming data. International Conference on Social Computing (SocialCom). Amsterdam : IEEE Computer Society, 2012. pp. 258-266
@inproceedings{63439f3b729144dbb7256b299a80bafb,
title = "Online sentiment and topic dynamics tracking over the streaming data",
abstract = "We propose a dynamic joint sentiment-topic model (dJST) which is able to effectively track sentiment and topic dynamics over the streaming data. Both topic and sentiment dynamics are captured by assuming that the current sentiment-topic specific word distributions are generated according to the word distributions at previous epochs. We study three different ways of accounting for such dependency information, (1) Sliding window where the current sentiment-topic-word distributions are dependent on the previous sentiment-topic specific word distributions in the last S epochs; (2) Skip model where history sentiment-topic-word distributions are considered by skipping some epochs in between; and (3) Multiscale model where previous long-and short-timescale distributions are taken into consideration. We derive efficient online inference procedures to sequentially update the model with newly arrived data and show the effectiveness of our proposed model on the Mozilla add-on reviews crawled between 2007 and 2011.",
author = "Chenghua Lin and Yulan He and Elizabeth Cano",
year = "2012",
doi = "10.1109/SocialCom-PASSAT.2012.14",
language = "English",
isbn = "978-1-4673-5638-1",
pages = "258--266",
booktitle = "International Conference on Social Computing (SocialCom)",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Online sentiment and topic dynamics tracking over the streaming data

AU - Lin, Chenghua

AU - He, Yulan

AU - Cano, Elizabeth

PY - 2012

Y1 - 2012

N2 - We propose a dynamic joint sentiment-topic model (dJST) which is able to effectively track sentiment and topic dynamics over the streaming data. Both topic and sentiment dynamics are captured by assuming that the current sentiment-topic specific word distributions are generated according to the word distributions at previous epochs. We study three different ways of accounting for such dependency information, (1) Sliding window where the current sentiment-topic-word distributions are dependent on the previous sentiment-topic specific word distributions in the last S epochs; (2) Skip model where history sentiment-topic-word distributions are considered by skipping some epochs in between; and (3) Multiscale model where previous long-and short-timescale distributions are taken into consideration. We derive efficient online inference procedures to sequentially update the model with newly arrived data and show the effectiveness of our proposed model on the Mozilla add-on reviews crawled between 2007 and 2011.

AB - We propose a dynamic joint sentiment-topic model (dJST) which is able to effectively track sentiment and topic dynamics over the streaming data. Both topic and sentiment dynamics are captured by assuming that the current sentiment-topic specific word distributions are generated according to the word distributions at previous epochs. We study three different ways of accounting for such dependency information, (1) Sliding window where the current sentiment-topic-word distributions are dependent on the previous sentiment-topic specific word distributions in the last S epochs; (2) Skip model where history sentiment-topic-word distributions are considered by skipping some epochs in between; and (3) Multiscale model where previous long-and short-timescale distributions are taken into consideration. We derive efficient online inference procedures to sequentially update the model with newly arrived data and show the effectiveness of our proposed model on the Mozilla add-on reviews crawled between 2007 and 2011.

U2 - 10.1109/SocialCom-PASSAT.2012.14

DO - 10.1109/SocialCom-PASSAT.2012.14

M3 - Conference contribution

SN - 978-1-4673-5638-1

SP - 258

EP - 266

BT - International Conference on Social Computing (SocialCom)

PB - IEEE Computer Society

CY - Amsterdam

ER -