Improving Tensor Based Recommenders with Clustering

Martin Leginus, Peter Dolog, Valdas Zemaitis

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

10 Citations (Scopus)

Abstract

Social tagging systems (STS) model three types of entities
(i.e. tag-user-item) and relationships between them are encoded into a
3-order tensor. Latent relationships and patterns can be discovered by
applying tensor factorization techniques like Higher Order Singular Value
Decomposition (HOSVD), Canonical Decomposition etc. STS accumulate large amount of sparse data that restricts factorization techniques
to detect latent relations and also significantly slows down the process
of a factorization. We propose to reduce tag space by exploiting clustering techniques so that the quality of the recommendations and execution time are improved and memory requirements are decreased. The
clustering is motivated by the fact that many tags in a tag space are
semantically similar thus the tags can be grouped. Finally, promising
experimental results are presented
Original languageEnglish
Title of host publicationUser Modeling, Adaptation, and Personalization : 20th International Conference, UMAP 2012, Montreal, Canada, July 16-20, 2012. Proceedings
Number of pages10
Volume7379
Place of PublicationBerlin
PublisherSpringer
Publication date2012
Pages151-163
ISBN (Print)978-3-642-31453-7
ISBN (Electronic)978-3-642-31454-4
DOIs
Publication statusPublished - 2012
EventUser Modeling, Adaptation, and Personalization - Montreal, Canada
Duration: 16 Jul 201220 Jul 2012
Conference number: 20

Conference

ConferenceUser Modeling, Adaptation, and Personalization
Number20
CountryCanada
CityMontreal
Period16/07/201220/07/2012
SeriesLecture Notes in Computer Science
Volume7379
ISSN0302-9743

Fingerprint

Factorization
Tensors
Decomposition
Data storage equipment

Keywords

  • tensor factorization
  • HOSVD
  • clustering.

Cite this

Leginus, M., Dolog, P., & Zemaitis, V. (2012). Improving Tensor Based Recommenders with Clustering. In User Modeling, Adaptation, and Personalization: 20th International Conference, UMAP 2012, Montreal, Canada, July 16-20, 2012. Proceedings (Vol. 7379, pp. 151-163). Berlin: Springer. Lecture Notes in Computer Science, Vol.. 7379 https://doi.org/10.1007/978-3-642-31454-4_13
Leginus, Martin ; Dolog, Peter ; Zemaitis, Valdas. / Improving Tensor Based Recommenders with Clustering. User Modeling, Adaptation, and Personalization: 20th International Conference, UMAP 2012, Montreal, Canada, July 16-20, 2012. Proceedings. Vol. 7379 Berlin : Springer, 2012. pp. 151-163 (Lecture Notes in Computer Science, Vol. 7379).
@inproceedings{36a8a0aafce44f8fb7e337f1b5c27e55,
title = "Improving Tensor Based Recommenders with Clustering",
abstract = "Social tagging systems (STS) model three types of entities(i.e. tag-user-item) and relationships between them are encoded into a3-order tensor. Latent relationships and patterns can be discovered byapplying tensor factorization techniques like Higher Order Singular ValueDecomposition (HOSVD), Canonical Decomposition etc. STS accumulate large amount of sparse data that restricts factorization techniquesto detect latent relations and also significantly slows down the processof a factorization. We propose to reduce tag space by exploiting clustering techniques so that the quality of the recommendations and execution time are improved and memory requirements are decreased. Theclustering is motivated by the fact that many tags in a tag space aresemantically similar thus the tags can be grouped. Finally, promisingexperimental results are presented",
keywords = "tensor factorization, HOSVD, clustering.",
author = "Martin Leginus and Peter Dolog and Valdas Zemaitis",
year = "2012",
doi = "10.1007/978-3-642-31454-4_13",
language = "English",
isbn = "978-3-642-31453-7",
volume = "7379",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "151--163",
booktitle = "User Modeling, Adaptation, and Personalization",
address = "Germany",

}

Leginus, M, Dolog, P & Zemaitis, V 2012, Improving Tensor Based Recommenders with Clustering. in User Modeling, Adaptation, and Personalization: 20th International Conference, UMAP 2012, Montreal, Canada, July 16-20, 2012. Proceedings. vol. 7379, Springer, Berlin, Lecture Notes in Computer Science, vol. 7379, pp. 151-163, User Modeling, Adaptation, and Personalization, Montreal, Canada, 16/07/2012. https://doi.org/10.1007/978-3-642-31454-4_13

Improving Tensor Based Recommenders with Clustering. / Leginus, Martin; Dolog, Peter; Zemaitis, Valdas.

User Modeling, Adaptation, and Personalization: 20th International Conference, UMAP 2012, Montreal, Canada, July 16-20, 2012. Proceedings. Vol. 7379 Berlin : Springer, 2012. p. 151-163 (Lecture Notes in Computer Science, Vol. 7379).

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Improving Tensor Based Recommenders with Clustering

AU - Leginus, Martin

AU - Dolog, Peter

AU - Zemaitis, Valdas

PY - 2012

Y1 - 2012

N2 - Social tagging systems (STS) model three types of entities(i.e. tag-user-item) and relationships between them are encoded into a3-order tensor. Latent relationships and patterns can be discovered byapplying tensor factorization techniques like Higher Order Singular ValueDecomposition (HOSVD), Canonical Decomposition etc. STS accumulate large amount of sparse data that restricts factorization techniquesto detect latent relations and also significantly slows down the processof a factorization. We propose to reduce tag space by exploiting clustering techniques so that the quality of the recommendations and execution time are improved and memory requirements are decreased. Theclustering is motivated by the fact that many tags in a tag space aresemantically similar thus the tags can be grouped. Finally, promisingexperimental results are presented

AB - Social tagging systems (STS) model three types of entities(i.e. tag-user-item) and relationships between them are encoded into a3-order tensor. Latent relationships and patterns can be discovered byapplying tensor factorization techniques like Higher Order Singular ValueDecomposition (HOSVD), Canonical Decomposition etc. STS accumulate large amount of sparse data that restricts factorization techniquesto detect latent relations and also significantly slows down the processof a factorization. We propose to reduce tag space by exploiting clustering techniques so that the quality of the recommendations and execution time are improved and memory requirements are decreased. Theclustering is motivated by the fact that many tags in a tag space aresemantically similar thus the tags can be grouped. Finally, promisingexperimental results are presented

KW - tensor factorization

KW - HOSVD

KW - clustering.

UR - http://www.scopus.com/inward/record.url?scp=84863614561&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-31454-4_13

DO - 10.1007/978-3-642-31454-4_13

M3 - Article in proceeding

SN - 978-3-642-31453-7

VL - 7379

T3 - Lecture Notes in Computer Science

SP - 151

EP - 163

BT - User Modeling, Adaptation, and Personalization

PB - Springer

CY - Berlin

ER -

Leginus M, Dolog P, Zemaitis V. Improving Tensor Based Recommenders with Clustering. In User Modeling, Adaptation, and Personalization: 20th International Conference, UMAP 2012, Montreal, Canada, July 16-20, 2012. Proceedings. Vol. 7379. Berlin: Springer. 2012. p. 151-163. (Lecture Notes in Computer Science, Vol. 7379). https://doi.org/10.1007/978-3-642-31454-4_13