Matrix factorization without user data retention

David Vallet, Arik Friedman, Shlomo Berkovsky

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionpeer-review

6 Citations (Scopus)


Recommender systems often rely on a centralized storage of user data and there is a growing concern about the misuse of the data. As recent studies have demonstrated, sensitive information could be inferred from user ratings stored by recommender systems. This work presents a novel semi-decentralized variant of the widely-used Matrix Factorization (MF) technique. The proposed approach eliminates the need for retaining user data, such that neither user ratings nor latent user vectors are stored by the recommender. Experimental evaluation indicates that the performance of the proposed approach is close to that of standard MF, and that the gap between the two diminishes as more user data becomes available. Our work paves the way to a new type of MF recommenders, which avoid user data retention, yet are capable of achieving accuracy similar to that of the state-of-the-art recommenders.
Original languageEnglish
Title of host publicationPacific-Asia Conference on Knowledge Discovery and Data Mining
Subtitle of host publication18th Pacific-Asia Conference, PAKDD 2014. Proceedings, Part I
EditorsVincent S. Tseng, Tu Bao Ho, Zhi-Hua Zhou, Arbee L. P. Chen, Hung-Yu Kao
Place of PublicationCham
PublisherSpringer, Springer Nature
Number of pages12
ISBN (Electronic)9783319066080
ISBN (Print)9783319066073
Publication statusPublished - 2014
Externally publishedYes
Event18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014 - Tainan, Taiwan
Duration: 13 May 201416 May 2014

Publication series

NameLecture Notes in Artificial Intelligence
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2014


  • data retention
  • matrix factorization
  • recommender systems


Dive into the research topics of 'Matrix factorization without user data retention'. Together they form a unique fingerprint.

Cite this