Boosting collaborative filtering based on missing data imputation using item's genre information

  • Weiwei Xia*
  • , Liang He
  • , Junzhong Gu
  • , Keqin He
  • , Lei Ren
  • *Corresponding author for this work

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

2 Scopus citations

Abstract

Collaborative filtering (CF) is one of the most successful technologies in recommender systems, and widely used in many personalized recommender applications, such as digital library, e-commerce, news sites, and so on. However, most collaborative filtering algorithms suffer from data sparsity problem which leads to inaccuracy of recommendation. This paper is with an eye to missing data imputation strategy in nearest-neighbor CF. We propose an effective CF framework based on missing data imputation before conducting CF process, which utilizes item's genre information. In the experimental evaluations, 19 item's genres are employed in the imputation stage. The results show that the proposed approaches effectively alleviate the negative impact of data sparsity, and perform better prediction accuracy than traditional widely-used CF algorithms.

Original languageEnglish
Title of host publicationProceedings - 2009 2nd IEEE International Conference on Computer Science and Information Technology, ICCSIT 2009
Pages332-336
Number of pages5
DOIs
StatePublished - 2009
Event2009 2nd IEEE International Conference on Computer Science and Information Technology, ICCSIT 2009 - Beijing, China
Duration: 8 Aug 200911 Aug 2009

Publication series

NameProceedings - 2009 2nd IEEE International Conference on Computer Science and Information Technology, ICCSIT 2009

Conference

Conference2009 2nd IEEE International Conference on Computer Science and Information Technology, ICCSIT 2009
Country/TerritoryChina
CityBeijing
Period8/08/0911/08/09

Keywords

  • Collaborative filtering
  • Missing data imputation
  • Recommender system
  • Sparsity problem

Fingerprint

Dive into the research topics of 'Boosting collaborative filtering based on missing data imputation using item's genre information'. Together they form a unique fingerprint.

Cite this