Exploring information hidden in tags: A subject-based item recommendation approach

Jing Peng, Dajun Zeng

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

5 Citations (Scopus)

Abstract

Collaborative tagging sites allow users to bookmark and annotate their favorite Web contents with tags. These tags provide a novel source of information for collaborative filtering (CF). Research on how to improve item recommendation quality leveraging tags is emerging yet information hidden in tags is far from being fully exploited. In this paper, we aim at finding informative usage patterns from tags by consistent clustering on tags using nonnegative matrix factorization. The clustered subjects, represented by weighed tag vectors, can then be used to build a subjectcentered user information seeking model for item recommendation. Experiments on two realworld datasets show that our subject-based algorithms substantially outperform the traditional CF methods as well as tag-enhanced recommendation approaches reported in the literature.

Original languageEnglish (US)
Title of host publication19th Workshop on Information Technologies and Systems, WITS 2009
PublisherSocial Science Research Network
Pages73-78
Number of pages6
StatePublished - 2009
Event19th Workshop on Information Technologies and Systems, WITS 2009 - Phoenix, AZ, United States
Duration: Dec 14 2009Dec 15 2009

Other

Other19th Workshop on Information Technologies and Systems, WITS 2009
CountryUnited States
CityPhoenix, AZ
Period12/14/0912/15/09

Fingerprint

Collaborative filtering
Factorization
Experiments

Keywords

  • Collaborative filtering
  • Collaborative tagging
  • Nonnegative matrix factorization
  • Tag-enhanced

ASJC Scopus subject areas

  • Information Systems
  • Control and Systems Engineering

Cite this

Peng, J., & Zeng, D. (2009). Exploring information hidden in tags: A subject-based item recommendation approach. In 19th Workshop on Information Technologies and Systems, WITS 2009 (pp. 73-78). Social Science Research Network.

Exploring information hidden in tags : A subject-based item recommendation approach. / Peng, Jing; Zeng, Dajun.

19th Workshop on Information Technologies and Systems, WITS 2009. Social Science Research Network, 2009. p. 73-78.

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

Peng, J & Zeng, D 2009, Exploring information hidden in tags: A subject-based item recommendation approach. in 19th Workshop on Information Technologies and Systems, WITS 2009. Social Science Research Network, pp. 73-78, 19th Workshop on Information Technologies and Systems, WITS 2009, Phoenix, AZ, United States, 12/14/09.
Peng J, Zeng D. Exploring information hidden in tags: A subject-based item recommendation approach. In 19th Workshop on Information Technologies and Systems, WITS 2009. Social Science Research Network. 2009. p. 73-78
Peng, Jing ; Zeng, Dajun. / Exploring information hidden in tags : A subject-based item recommendation approach. 19th Workshop on Information Technologies and Systems, WITS 2009. Social Science Research Network, 2009. pp. 73-78
@inproceedings{044fa0f7fd094256b4072fc1b511207a,
title = "Exploring information hidden in tags: A subject-based item recommendation approach",
abstract = "Collaborative tagging sites allow users to bookmark and annotate their favorite Web contents with tags. These tags provide a novel source of information for collaborative filtering (CF). Research on how to improve item recommendation quality leveraging tags is emerging yet information hidden in tags is far from being fully exploited. In this paper, we aim at finding informative usage patterns from tags by consistent clustering on tags using nonnegative matrix factorization. The clustered subjects, represented by weighed tag vectors, can then be used to build a subjectcentered user information seeking model for item recommendation. Experiments on two realworld datasets show that our subject-based algorithms substantially outperform the traditional CF methods as well as tag-enhanced recommendation approaches reported in the literature.",
keywords = "Collaborative filtering, Collaborative tagging, Nonnegative matrix factorization, Tag-enhanced",
author = "Jing Peng and Dajun Zeng",
year = "2009",
language = "English (US)",
pages = "73--78",
booktitle = "19th Workshop on Information Technologies and Systems, WITS 2009",
publisher = "Social Science Research Network",

}

TY - GEN

T1 - Exploring information hidden in tags

T2 - A subject-based item recommendation approach

AU - Peng, Jing

AU - Zeng, Dajun

PY - 2009

Y1 - 2009

N2 - Collaborative tagging sites allow users to bookmark and annotate their favorite Web contents with tags. These tags provide a novel source of information for collaborative filtering (CF). Research on how to improve item recommendation quality leveraging tags is emerging yet information hidden in tags is far from being fully exploited. In this paper, we aim at finding informative usage patterns from tags by consistent clustering on tags using nonnegative matrix factorization. The clustered subjects, represented by weighed tag vectors, can then be used to build a subjectcentered user information seeking model for item recommendation. Experiments on two realworld datasets show that our subject-based algorithms substantially outperform the traditional CF methods as well as tag-enhanced recommendation approaches reported in the literature.

AB - Collaborative tagging sites allow users to bookmark and annotate their favorite Web contents with tags. These tags provide a novel source of information for collaborative filtering (CF). Research on how to improve item recommendation quality leveraging tags is emerging yet information hidden in tags is far from being fully exploited. In this paper, we aim at finding informative usage patterns from tags by consistent clustering on tags using nonnegative matrix factorization. The clustered subjects, represented by weighed tag vectors, can then be used to build a subjectcentered user information seeking model for item recommendation. Experiments on two realworld datasets show that our subject-based algorithms substantially outperform the traditional CF methods as well as tag-enhanced recommendation approaches reported in the literature.

KW - Collaborative filtering

KW - Collaborative tagging

KW - Nonnegative matrix factorization

KW - Tag-enhanced

UR - http://www.scopus.com/inward/record.url?scp=84903838910&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84903838910&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84903838910

SP - 73

EP - 78

BT - 19th Workshop on Information Technologies and Systems, WITS 2009

PB - Social Science Research Network

ER -