Scalable Preference Learning from Data Streams

Fabon Dzogang; Thomas Lansdall-Welfare; Saatviga Sudhahar; Nello Cristianini

doi:10.1145/2740908.2742008

Scalable Preference Learning from Data Streams

Fabon Dzogang, Thomas Lansdall-Welfare, Saatviga Sudhahar, Nello Cristianini

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

7 Citations (Scopus)

Abstract

We study the task of learning the preferences of online readers of news, based on their past choices. Previous work has shown that it is possible to model this situation as a competition between articles, where the most appealing articles of the day are those selected by the most users. The appeal of an article can be computed from its textual content, and the evaluation function can be learned from training data. In this paper, we show how this task can benefit from an efficient algorithm, based on hashing representations, which enables it to be deployed on high intensity data streams. We demonstrate the effectiveness of this approach on four real world news streams, compare it with standard approaches, and describe a new online demonstration based on this technology.

Original language	English
Title of host publication	WWW '15 Companion Proceedings of the 24th International Conference on World Wide Web
Publisher	Association for Computing Machinery (ACM)
Pages	885-890
ISBN (Print)	978-1-4503-3473-0
DOIs	https://doi.org/10.1145/2740908.2742008
Publication status	Published - May 2015
Event	NewsWWW 2015 – 2nd workshop on Web and Data Science for News Publishing - WWW'15, Florence, Italy Duration: 18 May 2015 → 22 May 2015

Conference

Conference	NewsWWW 2015 – 2nd workshop on Web and Data Science for News Publishing
Country/Territory	Italy
City	Florence
Period	18/05/15 → 22/05/15

Keywords

News popularity
Online learning
News appeal
Learning to rank
Hashing trick

Access to Document

10.1145/2740908.2742008

Handle.net

Persistent link

Cite this

@inproceedings{a737a08458404f60884a744824658df9,

title = "Scalable Preference Learning from Data Streams",

abstract = "We study the task of learning the preferences of online readers of news, based on their past choices. Previous work has shown that it is possible to model this situation as a competition between articles, where the most appealing articles of the day are those selected by the most users. The appeal of an article can be computed from its textual content, and the evaluation function can be learned from training data. In this paper, we show how this task can benefit from an efficient algorithm, based on hashing representations, which enables it to be deployed on high intensity data streams. We demonstrate the effectiveness of this approach on four real world news streams, compare it with standard approaches, and describe a new online demonstration based on this technology.",

keywords = "News popularity, Online learning, News appeal, Learning to rank, Hashing trick",

author = "Fabon Dzogang and Thomas Lansdall-Welfare and Saatviga Sudhahar and Nello Cristianini",

year = "2015",

month = may,

doi = "10.1145/2740908.2742008",

language = "English",

isbn = "978-1-4503-3473-0",

pages = "885--890",

booktitle = "WWW '15 Companion Proceedings of the 24th International Conference on World Wide Web",

publisher = "Association for Computing Machinery (ACM)",

address = "United States",

note = "NewsWWW 2015 – 2nd workshop on Web and Data Science for News Publishing ; Conference date: 18-05-2015 Through 22-05-2015",

}

Dzogang, F, Lansdall-Welfare, T, Sudhahar, S & Cristianini, N 2015, Scalable Preference Learning from Data Streams. in WWW '15 Companion Proceedings of the 24th International Conference on World Wide Web. Association for Computing Machinery (ACM), pp. 885-890, NewsWWW 2015 – 2nd workshop on Web and Data Science for News Publishing, Florence, Italy, 18/05/15. https://doi.org/10.1145/2740908.2742008

Scalable Preference Learning from Data Streams. / Dzogang, Fabon; Lansdall-Welfare, Thomas; Sudhahar, Saatviga et al.
WWW '15 Companion Proceedings of the 24th International Conference on World Wide Web. Association for Computing Machinery (ACM), 2015. p. 885-890.

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

TY - GEN

T1 - Scalable Preference Learning from Data Streams

AU - Dzogang, Fabon

AU - Lansdall-Welfare, Thomas

AU - Sudhahar, Saatviga

AU - Cristianini, Nello

PY - 2015/5

Y1 - 2015/5

N2 - We study the task of learning the preferences of online readers of news, based on their past choices. Previous work has shown that it is possible to model this situation as a competition between articles, where the most appealing articles of the day are those selected by the most users. The appeal of an article can be computed from its textual content, and the evaluation function can be learned from training data. In this paper, we show how this task can benefit from an efficient algorithm, based on hashing representations, which enables it to be deployed on high intensity data streams. We demonstrate the effectiveness of this approach on four real world news streams, compare it with standard approaches, and describe a new online demonstration based on this technology.

AB - We study the task of learning the preferences of online readers of news, based on their past choices. Previous work has shown that it is possible to model this situation as a competition between articles, where the most appealing articles of the day are those selected by the most users. The appeal of an article can be computed from its textual content, and the evaluation function can be learned from training data. In this paper, we show how this task can benefit from an efficient algorithm, based on hashing representations, which enables it to be deployed on high intensity data streams. We demonstrate the effectiveness of this approach on four real world news streams, compare it with standard approaches, and describe a new online demonstration based on this technology.

KW - News popularity

KW - Online learning

KW - News appeal

KW - Learning to rank

KW - Hashing trick

U2 - 10.1145/2740908.2742008

DO - 10.1145/2740908.2742008

M3 - Conference Contribution (Conference Proceeding)

SN - 978-1-4503-3473-0

SP - 885

EP - 890

BT - WWW '15 Companion Proceedings of the 24th International Conference on World Wide Web

PB - Association for Computing Machinery (ACM)

T2 - NewsWWW 2015 – 2nd workshop on Web and Data Science for News Publishing

Y2 - 18 May 2015 through 22 May 2015

ER -

Scalable Preference Learning from Data Streams

Abstract

Conference

Keywords

Access to Document

Handle.net

Fingerprint

Cite this