Recommendations over domain specific user graphs

Makoto Nakatsuji, Yasuhiro Fujiwara, Akimichi Tanaka, Tadasu Uchiyama, Toru Ishida

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

14 Citations (Scopus)

Abstract

Content providers want to make recommendations across multiple interrelated domains such as music and movies. However, existing collaborative filtering methods fail to accurately identify items that may be interesting to the user but that lie in domains that the user has not accessed before. This is mainly because of the paucity of user transactions across multiple item domains. Our method is based on the observation that users who share similar items or who share social connections, can provide recommendation chains (sequences of transitively associated edges) to items in other domains. It first builds domain-specific-usergraphs (DSUGs) whose nodes, users, are linked by weighted edges that reflect user similarity. It then connects the DSUGs via the users who rated items in several domains or via the users who share social connections, to create a cross-domain-user graph (CDUG). It performs Random Walk with Restarts on the CDUG to extract user nodes that are related to the starting user node on the CDUG even though they are not present in the DSUG of the starting user node. It then adds items possessed by those users to the recommendations of the starting node user. Furthermore, to extract many more user nodes, we employ a taxonomy-based similarity measure that states that users are similar if they share the same items and/or same classes. Thus we can set many suitable routes from the starting user node to other user nodes in the CDUG. An evaluation using rating datasets in two interrelated domains and social connection histories of users as extracted from a blog portal, indicates that our method identifies potentially interesting items in other domains with higher accuracy than is possible with existing CF methods.

Original languageEnglish
Title of host publicationECAI 2010
PublisherIOS Press
Pages607-612
Number of pages6
ISBN (Print)9781607506058
DOIs
Publication statusPublished - 2010 Jan 1
Externally publishedYes
Event2nd Workshop on Knowledge Representation for Health Care, KR4HC 2010, held in conjunction with the 19th European Conference in Artificial Intelligence, ECAI 2010 - Lisbon, Portugal
Duration: 2010 Aug 172010 Aug 17

Publication series

NameFrontiers in Artificial Intelligence and Applications
Volume215
ISSN (Print)0922-6389

Conference

Conference2nd Workshop on Knowledge Representation for Health Care, KR4HC 2010, held in conjunction with the 19th European Conference in Artificial Intelligence, ECAI 2010
CountryPortugal
CityLisbon
Period10/8/1710/8/17

Fingerprint

Collaborative filtering
Blogs
Taxonomies

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Nakatsuji, M., Fujiwara, Y., Tanaka, A., Uchiyama, T., & Ishida, T. (2010). Recommendations over domain specific user graphs. In ECAI 2010 (pp. 607-612). (Frontiers in Artificial Intelligence and Applications; Vol. 215). IOS Press. https://doi.org/10.3233/978-1-60750-606-5-607

Recommendations over domain specific user graphs. / Nakatsuji, Makoto; Fujiwara, Yasuhiro; Tanaka, Akimichi; Uchiyama, Tadasu; Ishida, Toru.

ECAI 2010. IOS Press, 2010. p. 607-612 (Frontiers in Artificial Intelligence and Applications; Vol. 215).

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

Nakatsuji, M, Fujiwara, Y, Tanaka, A, Uchiyama, T & Ishida, T 2010, Recommendations over domain specific user graphs. in ECAI 2010. Frontiers in Artificial Intelligence and Applications, vol. 215, IOS Press, pp. 607-612, 2nd Workshop on Knowledge Representation for Health Care, KR4HC 2010, held in conjunction with the 19th European Conference in Artificial Intelligence, ECAI 2010, Lisbon, Portugal, 10/8/17. https://doi.org/10.3233/978-1-60750-606-5-607
Nakatsuji M, Fujiwara Y, Tanaka A, Uchiyama T, Ishida T. Recommendations over domain specific user graphs. In ECAI 2010. IOS Press. 2010. p. 607-612. (Frontiers in Artificial Intelligence and Applications). https://doi.org/10.3233/978-1-60750-606-5-607
Nakatsuji, Makoto ; Fujiwara, Yasuhiro ; Tanaka, Akimichi ; Uchiyama, Tadasu ; Ishida, Toru. / Recommendations over domain specific user graphs. ECAI 2010. IOS Press, 2010. pp. 607-612 (Frontiers in Artificial Intelligence and Applications).
@inproceedings{b139b599be24466e9e628120091c718c,
title = "Recommendations over domain specific user graphs",
abstract = "Content providers want to make recommendations across multiple interrelated domains such as music and movies. However, existing collaborative filtering methods fail to accurately identify items that may be interesting to the user but that lie in domains that the user has not accessed before. This is mainly because of the paucity of user transactions across multiple item domains. Our method is based on the observation that users who share similar items or who share social connections, can provide recommendation chains (sequences of transitively associated edges) to items in other domains. It first builds domain-specific-usergraphs (DSUGs) whose nodes, users, are linked by weighted edges that reflect user similarity. It then connects the DSUGs via the users who rated items in several domains or via the users who share social connections, to create a cross-domain-user graph (CDUG). It performs Random Walk with Restarts on the CDUG to extract user nodes that are related to the starting user node on the CDUG even though they are not present in the DSUG of the starting user node. It then adds items possessed by those users to the recommendations of the starting node user. Furthermore, to extract many more user nodes, we employ a taxonomy-based similarity measure that states that users are similar if they share the same items and/or same classes. Thus we can set many suitable routes from the starting user node to other user nodes in the CDUG. An evaluation using rating datasets in two interrelated domains and social connection histories of users as extracted from a blog portal, indicates that our method identifies potentially interesting items in other domains with higher accuracy than is possible with existing CF methods.",
author = "Makoto Nakatsuji and Yasuhiro Fujiwara and Akimichi Tanaka and Tadasu Uchiyama and Toru Ishida",
year = "2010",
month = "1",
day = "1",
doi = "10.3233/978-1-60750-606-5-607",
language = "English",
isbn = "9781607506058",
series = "Frontiers in Artificial Intelligence and Applications",
publisher = "IOS Press",
pages = "607--612",
booktitle = "ECAI 2010",
address = "Netherlands",

}

TY - GEN

T1 - Recommendations over domain specific user graphs

AU - Nakatsuji, Makoto

AU - Fujiwara, Yasuhiro

AU - Tanaka, Akimichi

AU - Uchiyama, Tadasu

AU - Ishida, Toru

PY - 2010/1/1

Y1 - 2010/1/1

N2 - Content providers want to make recommendations across multiple interrelated domains such as music and movies. However, existing collaborative filtering methods fail to accurately identify items that may be interesting to the user but that lie in domains that the user has not accessed before. This is mainly because of the paucity of user transactions across multiple item domains. Our method is based on the observation that users who share similar items or who share social connections, can provide recommendation chains (sequences of transitively associated edges) to items in other domains. It first builds domain-specific-usergraphs (DSUGs) whose nodes, users, are linked by weighted edges that reflect user similarity. It then connects the DSUGs via the users who rated items in several domains or via the users who share social connections, to create a cross-domain-user graph (CDUG). It performs Random Walk with Restarts on the CDUG to extract user nodes that are related to the starting user node on the CDUG even though they are not present in the DSUG of the starting user node. It then adds items possessed by those users to the recommendations of the starting node user. Furthermore, to extract many more user nodes, we employ a taxonomy-based similarity measure that states that users are similar if they share the same items and/or same classes. Thus we can set many suitable routes from the starting user node to other user nodes in the CDUG. An evaluation using rating datasets in two interrelated domains and social connection histories of users as extracted from a blog portal, indicates that our method identifies potentially interesting items in other domains with higher accuracy than is possible with existing CF methods.

AB - Content providers want to make recommendations across multiple interrelated domains such as music and movies. However, existing collaborative filtering methods fail to accurately identify items that may be interesting to the user but that lie in domains that the user has not accessed before. This is mainly because of the paucity of user transactions across multiple item domains. Our method is based on the observation that users who share similar items or who share social connections, can provide recommendation chains (sequences of transitively associated edges) to items in other domains. It first builds domain-specific-usergraphs (DSUGs) whose nodes, users, are linked by weighted edges that reflect user similarity. It then connects the DSUGs via the users who rated items in several domains or via the users who share social connections, to create a cross-domain-user graph (CDUG). It performs Random Walk with Restarts on the CDUG to extract user nodes that are related to the starting user node on the CDUG even though they are not present in the DSUG of the starting user node. It then adds items possessed by those users to the recommendations of the starting node user. Furthermore, to extract many more user nodes, we employ a taxonomy-based similarity measure that states that users are similar if they share the same items and/or same classes. Thus we can set many suitable routes from the starting user node to other user nodes in the CDUG. An evaluation using rating datasets in two interrelated domains and social connection histories of users as extracted from a blog portal, indicates that our method identifies potentially interesting items in other domains with higher accuracy than is possible with existing CF methods.

UR - http://www.scopus.com/inward/record.url?scp=77956049387&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956049387&partnerID=8YFLogxK

U2 - 10.3233/978-1-60750-606-5-607

DO - 10.3233/978-1-60750-606-5-607

M3 - Conference contribution

SN - 9781607506058

T3 - Frontiers in Artificial Intelligence and Applications

SP - 607

EP - 612

BT - ECAI 2010

PB - IOS Press

ER -