L-Injection: Toward effective collaborative filtering using uninteresting items

Jongwuk Lee, Won Seok Hwang, Juan Parc, Youngnam Lee, Sang Wook Kim, Dongwon Lee

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

We develop a novel framework, named as l-injection, to address the sparsity problem of recommender systems. By carefully injecting low values to a selected set of unrated user-item pairs in a user-item matrix, we demonstrate that top-N recommendation accuracies of various collaborative filtering (CF) techniques can be significantly and consistently improved. We first adopt the notion of pre-use preferences of users toward a vast amount of unrated items. Using this notion, we identify uninteresting items that have not been rated yet but are likely to receive low ratings from users, and selectively impute them as low values. As our proposed approach is method-Agnostic, it can be easily applied to a variety of CF algorithms. Through comprehensive experiments with three real-life datasets (e.g., Movielens, Ciao, and Watcha), we demonstrate that our solution consistently and universally enhances the accuracies of existing CF algorithms (e.g., item-based CF, SVD-based CF, and SVD++) by 2.5 to 5 times on average. Furthermore, our solution improves the running time of those CF methods by 1.2 to 2.3 times when its setting produces the best accuracy. The datasets and codes that we used in the experiments are available at: https://goo.gl/KUrmip.

Original languageEnglish (US)
Article number7913668
Pages (from-to)3-16
Number of pages14
JournalIEEE Transactions on Knowledge and Data Engineering
Volume31
Issue number1
DOIs
StatePublished - Jan 1 2019

Fingerprint

Collaborative filtering
Singular value decomposition
Recommender systems
Experiments

All Science Journal Classification (ASJC) codes

  • Information Systems
  • Computer Science Applications
  • Computational Theory and Mathematics

Cite this

Lee, Jongwuk ; Hwang, Won Seok ; Parc, Juan ; Lee, Youngnam ; Kim, Sang Wook ; Lee, Dongwon. / L-Injection : Toward effective collaborative filtering using uninteresting items. In: IEEE Transactions on Knowledge and Data Engineering. 2019 ; Vol. 31, No. 1. pp. 3-16.
@article{d0718d305d084cfc84f574059423820d,
title = "L-Injection: Toward effective collaborative filtering using uninteresting items",
abstract = "We develop a novel framework, named as l-injection, to address the sparsity problem of recommender systems. By carefully injecting low values to a selected set of unrated user-item pairs in a user-item matrix, we demonstrate that top-N recommendation accuracies of various collaborative filtering (CF) techniques can be significantly and consistently improved. We first adopt the notion of pre-use preferences of users toward a vast amount of unrated items. Using this notion, we identify uninteresting items that have not been rated yet but are likely to receive low ratings from users, and selectively impute them as low values. As our proposed approach is method-Agnostic, it can be easily applied to a variety of CF algorithms. Through comprehensive experiments with three real-life datasets (e.g., Movielens, Ciao, and Watcha), we demonstrate that our solution consistently and universally enhances the accuracies of existing CF algorithms (e.g., item-based CF, SVD-based CF, and SVD++) by 2.5 to 5 times on average. Furthermore, our solution improves the running time of those CF methods by 1.2 to 2.3 times when its setting produces the best accuracy. The datasets and codes that we used in the experiments are available at: https://goo.gl/KUrmip.",
author = "Jongwuk Lee and Hwang, {Won Seok} and Juan Parc and Youngnam Lee and Kim, {Sang Wook} and Dongwon Lee",
year = "2019",
month = "1",
day = "1",
doi = "10.1109/TKDE.2017.2698461",
language = "English (US)",
volume = "31",
pages = "3--16",
journal = "IEEE Transactions on Knowledge and Data Engineering",
issn = "1041-4347",
publisher = "IEEE Computer Society",
number = "1",

}

L-Injection : Toward effective collaborative filtering using uninteresting items. / Lee, Jongwuk; Hwang, Won Seok; Parc, Juan; Lee, Youngnam; Kim, Sang Wook; Lee, Dongwon.

In: IEEE Transactions on Knowledge and Data Engineering, Vol. 31, No. 1, 7913668, 01.01.2019, p. 3-16.

Research output: Contribution to journalArticle

TY - JOUR

T1 - L-Injection

T2 - Toward effective collaborative filtering using uninteresting items

AU - Lee, Jongwuk

AU - Hwang, Won Seok

AU - Parc, Juan

AU - Lee, Youngnam

AU - Kim, Sang Wook

AU - Lee, Dongwon

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We develop a novel framework, named as l-injection, to address the sparsity problem of recommender systems. By carefully injecting low values to a selected set of unrated user-item pairs in a user-item matrix, we demonstrate that top-N recommendation accuracies of various collaborative filtering (CF) techniques can be significantly and consistently improved. We first adopt the notion of pre-use preferences of users toward a vast amount of unrated items. Using this notion, we identify uninteresting items that have not been rated yet but are likely to receive low ratings from users, and selectively impute them as low values. As our proposed approach is method-Agnostic, it can be easily applied to a variety of CF algorithms. Through comprehensive experiments with three real-life datasets (e.g., Movielens, Ciao, and Watcha), we demonstrate that our solution consistently and universally enhances the accuracies of existing CF algorithms (e.g., item-based CF, SVD-based CF, and SVD++) by 2.5 to 5 times on average. Furthermore, our solution improves the running time of those CF methods by 1.2 to 2.3 times when its setting produces the best accuracy. The datasets and codes that we used in the experiments are available at: https://goo.gl/KUrmip.

AB - We develop a novel framework, named as l-injection, to address the sparsity problem of recommender systems. By carefully injecting low values to a selected set of unrated user-item pairs in a user-item matrix, we demonstrate that top-N recommendation accuracies of various collaborative filtering (CF) techniques can be significantly and consistently improved. We first adopt the notion of pre-use preferences of users toward a vast amount of unrated items. Using this notion, we identify uninteresting items that have not been rated yet but are likely to receive low ratings from users, and selectively impute them as low values. As our proposed approach is method-Agnostic, it can be easily applied to a variety of CF algorithms. Through comprehensive experiments with three real-life datasets (e.g., Movielens, Ciao, and Watcha), we demonstrate that our solution consistently and universally enhances the accuracies of existing CF algorithms (e.g., item-based CF, SVD-based CF, and SVD++) by 2.5 to 5 times on average. Furthermore, our solution improves the running time of those CF methods by 1.2 to 2.3 times when its setting produces the best accuracy. The datasets and codes that we used in the experiments are available at: https://goo.gl/KUrmip.

UR - http://www.scopus.com/inward/record.url?scp=85058210716&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058210716&partnerID=8YFLogxK

U2 - 10.1109/TKDE.2017.2698461

DO - 10.1109/TKDE.2017.2698461

M3 - Article

AN - SCOPUS:85058210716

VL - 31

SP - 3

EP - 16

JO - IEEE Transactions on Knowledge and Data Engineering

JF - IEEE Transactions on Knowledge and Data Engineering

SN - 1041-4347

IS - 1

M1 - 7913668

ER -