Efficient Rectangular Maximal-Volume Algorithm for Rating Elicitation in Collaborative Filtering

Handle URI:
http://hdl.handle.net/10754/623827
Title:
Efficient Rectangular Maximal-Volume Algorithm for Rating Elicitation in Collaborative Filtering
Authors:
Fonarev, Alexander; Mikhalev, Alexander; Serdyukov, Pavel; Gusev, Gleb; Oseledets, Ivan
Abstract:
Cold start problem in Collaborative Filtering can be solved by asking new users to rate a small seed set of representative items or by asking representative users to rate a new item. The question is how to build a seed set that can give enough preference information for making good recommendations. One of the most successful approaches, called Representative Based Matrix Factorization, is based on Maxvol algorithm. Unfortunately, this approach has one important limitation - a seed set of a particular size requires a rating matrix factorization of fixed rank that should coincide with that size. This is not necessarily optimal in the general case. In the current paper, we introduce a fast algorithm for an analytical generalization of this approach that we call Rectangular Maxvol. It allows the rank of factorization to be lower than the required size of the seed set. Moreover, the paper includes the theoretical analysis of the method's error, the complexity analysis of the existing methods and the comparison to the state-of-the-art approaches.
KAUST Department:
King Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabia
Citation:
Fonarev A, Mikhalev A, Serdyukov P, Gusev G, Oseledets I (2016) Efficient Rectangular Maximal-Volume Algorithm for Rating Elicitation in Collaborative Filtering. 2016 IEEE 16th International Conference on Data Mining (ICDM). Available: http://dx.doi.org/10.1109/ICDM.2016.0025.
Publisher:
Institute of Electrical and Electronics Engineers (IEEE)
Journal:
2016 IEEE 16th International Conference on Data Mining (ICDM)
Issue Date:
7-Feb-2017
DOI:
10.1109/ICDM.2016.0025
Type:
Conference Paper
Sponsors:
Work on problem setting and numerical examples was supported by Russian Science Foundation grant 14-11-00659. Work on theoretical estimations of approximation error and practical algorithm was supported by Russian Foundation for Basic Research 16-31-00351 mol_a. Also we thank Evgeny Frolov for helpful discussions.
Additional Links:
http://ieeexplore.ieee.org/document/7837838/
Appears in Collections:
Conference Papers

Full metadata record

DC FieldValue Language
dc.contributor.authorFonarev, Alexanderen
dc.contributor.authorMikhalev, Alexanderen
dc.contributor.authorSerdyukov, Pavelen
dc.contributor.authorGusev, Gleben
dc.contributor.authorOseledets, Ivanen
dc.date.accessioned2017-05-31T11:23:08Z-
dc.date.available2017-05-31T11:23:08Z-
dc.date.issued2017-02-07en
dc.identifier.citationFonarev A, Mikhalev A, Serdyukov P, Gusev G, Oseledets I (2016) Efficient Rectangular Maximal-Volume Algorithm for Rating Elicitation in Collaborative Filtering. 2016 IEEE 16th International Conference on Data Mining (ICDM). Available: http://dx.doi.org/10.1109/ICDM.2016.0025.en
dc.identifier.doi10.1109/ICDM.2016.0025en
dc.identifier.urihttp://hdl.handle.net/10754/623827-
dc.description.abstractCold start problem in Collaborative Filtering can be solved by asking new users to rate a small seed set of representative items or by asking representative users to rate a new item. The question is how to build a seed set that can give enough preference information for making good recommendations. One of the most successful approaches, called Representative Based Matrix Factorization, is based on Maxvol algorithm. Unfortunately, this approach has one important limitation - a seed set of a particular size requires a rating matrix factorization of fixed rank that should coincide with that size. This is not necessarily optimal in the general case. In the current paper, we introduce a fast algorithm for an analytical generalization of this approach that we call Rectangular Maxvol. It allows the rank of factorization to be lower than the required size of the seed set. Moreover, the paper includes the theoretical analysis of the method's error, the complexity analysis of the existing methods and the comparison to the state-of-the-art approaches.en
dc.description.sponsorshipWork on problem setting and numerical examples was supported by Russian Science Foundation grant 14-11-00659. Work on theoretical estimations of approximation error and practical algorithm was supported by Russian Foundation for Basic Research 16-31-00351 mol_a. Also we thank Evgeny Frolov for helpful discussions.en
dc.publisherInstitute of Electrical and Electronics Engineers (IEEE)en
dc.relation.urlhttp://ieeexplore.ieee.org/document/7837838/en
dc.subjectcollaborative filteringen
dc.subjectmatrix decompositionen
dc.subjectAlgorithm design and analysisen
dc.subjectCollaborationen
dc.subjectFilteringen
dc.subjectMathematical modelen
dc.subjectMatrix decompositionen
dc.subjectOptimizationen
dc.subjectPrediction algorithmsen
dc.titleEfficient Rectangular Maximal-Volume Algorithm for Rating Elicitation in Collaborative Filteringen
dc.typeConference Paperen
dc.contributor.departmentKing Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabiaen
dc.identifier.journal2016 IEEE 16th International Conference on Data Mining (ICDM)en
dc.contributor.institutionSkolkovo Institute of Science and Technology, Moscow, Russiaen
dc.contributor.institutionYandex, Moscow, Russiaen
dc.contributor.institutionSBDA Group, Dublin, Irelanden
dc.contributor.institutionInstitute of Numerical Mathematics, Russian Academy of Sciences, Moscow, Russiaen
kaust.authorMikhalev, Alexanderen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.