Recommendation as link prediction

A graph kernel-based machine learning approach

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

22 Citations (Scopus)

Abstract

Recommender systems have demonstrated commercial success in multiple industries. In digital libraries they have the potential to be used as a support tool for traditional information retrieval functions. Among the major recommendation algorithms, the successful collaborative filtering (CF) methods explore the use of user-item interactions to infer user interests. Based on the finding that transitive user-item associations can alleviate the data sparsity problem in CF, multiple heuristic algorithms were designed to take advantage of the user-item interaction networks with both direct and indirect interactions. However, the use of such graph representation was still limited in learning-based algorithms. In this paper, we propose a graph kernel-based recommendation framework. For each user-item pair, we inspect its associative interaction graph (AIG) that contains the users, items, and interactions n steps away from the pair. We design a novel graph kernel to capture the AIG structures and use them to predict possible user-item interactions. The framework demonstrates improved performance on an online bookstore dataset, especially when a large number of suggestions are needed.

Original languageEnglish (US)
Title of host publicationProceedings of the ACM/IEEE Joint Conference on Digital Libraries
Pages213-216
Number of pages4
DOIs
StatePublished - 2009
Externally publishedYes
Event2009 ACM/IEEE Joint Conference on Digital Libraries, JCDL'09 - Austin, TX, United States
Duration: Jun 15 2009Jun 19 2009

Other

Other2009 ACM/IEEE Joint Conference on Digital Libraries, JCDL'09
CountryUnited States
CityAustin, TX
Period6/15/096/19/09

Fingerprint

Collaborative filtering
Learning systems
Digital libraries
Recommender systems
Heuristic algorithms
Information retrieval
Industry

Keywords

  • Collaborative filtering
  • Kernel methods
  • Recommender system

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Li, X., & Chen, H. (2009). Recommendation as link prediction: A graph kernel-based machine learning approach. In Proceedings of the ACM/IEEE Joint Conference on Digital Libraries (pp. 213-216) https://doi.org/10.1145/1555400.1555433

Recommendation as link prediction : A graph kernel-based machine learning approach. / Li, Xin; Chen, Hsinchun.

Proceedings of the ACM/IEEE Joint Conference on Digital Libraries. 2009. p. 213-216.

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

Li, X & Chen, H 2009, Recommendation as link prediction: A graph kernel-based machine learning approach. in Proceedings of the ACM/IEEE Joint Conference on Digital Libraries. pp. 213-216, 2009 ACM/IEEE Joint Conference on Digital Libraries, JCDL'09, Austin, TX, United States, 6/15/09. https://doi.org/10.1145/1555400.1555433
Li X, Chen H. Recommendation as link prediction: A graph kernel-based machine learning approach. In Proceedings of the ACM/IEEE Joint Conference on Digital Libraries. 2009. p. 213-216 https://doi.org/10.1145/1555400.1555433
Li, Xin ; Chen, Hsinchun. / Recommendation as link prediction : A graph kernel-based machine learning approach. Proceedings of the ACM/IEEE Joint Conference on Digital Libraries. 2009. pp. 213-216
@inproceedings{300bc7c205aa4976b35a9f891679e272,
title = "Recommendation as link prediction: A graph kernel-based machine learning approach",
abstract = "Recommender systems have demonstrated commercial success in multiple industries. In digital libraries they have the potential to be used as a support tool for traditional information retrieval functions. Among the major recommendation algorithms, the successful collaborative filtering (CF) methods explore the use of user-item interactions to infer user interests. Based on the finding that transitive user-item associations can alleviate the data sparsity problem in CF, multiple heuristic algorithms were designed to take advantage of the user-item interaction networks with both direct and indirect interactions. However, the use of such graph representation was still limited in learning-based algorithms. In this paper, we propose a graph kernel-based recommendation framework. For each user-item pair, we inspect its associative interaction graph (AIG) that contains the users, items, and interactions n steps away from the pair. We design a novel graph kernel to capture the AIG structures and use them to predict possible user-item interactions. The framework demonstrates improved performance on an online bookstore dataset, especially when a large number of suggestions are needed.",
keywords = "Collaborative filtering, Kernel methods, Recommender system",
author = "Xin Li and Hsinchun Chen",
year = "2009",
doi = "10.1145/1555400.1555433",
language = "English (US)",
isbn = "9781605586977",
pages = "213--216",
booktitle = "Proceedings of the ACM/IEEE Joint Conference on Digital Libraries",

}

TY - GEN

T1 - Recommendation as link prediction

T2 - A graph kernel-based machine learning approach

AU - Li, Xin

AU - Chen, Hsinchun

PY - 2009

Y1 - 2009

N2 - Recommender systems have demonstrated commercial success in multiple industries. In digital libraries they have the potential to be used as a support tool for traditional information retrieval functions. Among the major recommendation algorithms, the successful collaborative filtering (CF) methods explore the use of user-item interactions to infer user interests. Based on the finding that transitive user-item associations can alleviate the data sparsity problem in CF, multiple heuristic algorithms were designed to take advantage of the user-item interaction networks with both direct and indirect interactions. However, the use of such graph representation was still limited in learning-based algorithms. In this paper, we propose a graph kernel-based recommendation framework. For each user-item pair, we inspect its associative interaction graph (AIG) that contains the users, items, and interactions n steps away from the pair. We design a novel graph kernel to capture the AIG structures and use them to predict possible user-item interactions. The framework demonstrates improved performance on an online bookstore dataset, especially when a large number of suggestions are needed.

AB - Recommender systems have demonstrated commercial success in multiple industries. In digital libraries they have the potential to be used as a support tool for traditional information retrieval functions. Among the major recommendation algorithms, the successful collaborative filtering (CF) methods explore the use of user-item interactions to infer user interests. Based on the finding that transitive user-item associations can alleviate the data sparsity problem in CF, multiple heuristic algorithms were designed to take advantage of the user-item interaction networks with both direct and indirect interactions. However, the use of such graph representation was still limited in learning-based algorithms. In this paper, we propose a graph kernel-based recommendation framework. For each user-item pair, we inspect its associative interaction graph (AIG) that contains the users, items, and interactions n steps away from the pair. We design a novel graph kernel to capture the AIG structures and use them to predict possible user-item interactions. The framework demonstrates improved performance on an online bookstore dataset, especially when a large number of suggestions are needed.

KW - Collaborative filtering

KW - Kernel methods

KW - Recommender system

UR - http://www.scopus.com/inward/record.url?scp=70450235024&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70450235024&partnerID=8YFLogxK

U2 - 10.1145/1555400.1555433

DO - 10.1145/1555400.1555433

M3 - Conference contribution

SN - 9781605586977

SP - 213

EP - 216

BT - Proceedings of the ACM/IEEE Joint Conference on Digital Libraries

ER -