Shilling attack detection in recommender systems via selecting patterns analysis

Wentao Li, Min Gao, Hua Li, Jun Zeng, Qingyu Xiong, Sachio Hirokawa

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Collaborative filtering (CF) has been widely used in recommender systems to generate personalized recommendations. However, recommender systems using CF are vulnerable to shilling attacks, in which attackers inject fake profiles to manipulate recommendation results. Thus, shilling attacks pose a threat to the credibility of recommender systems. Previous studies mainly derive features from characteristics of item ratings in user profiles to detect attackers, but the methods suffer from low accuracy when attackers adopt new rating patterns. To overcome this drawback, we derive features from properties of item popularity in user profiles, which are determined by users' different selecting patterns. This feature extraction method is based on the prior knowledge that attackers select items to rate with man-made rules while normal users do this according to their inner preferences. Then, machine learning classification approaches are exploited to make use of these features to detect and remove attackers. Experiment results on the MovieLens dataset and Amazon review dataset show that our proposed method improves detection performance. In addition, the results justify the practical value of features derived from selecting patterns.

Original languageEnglish
Pages (from-to)2600-2611
Number of pages12
JournalIEICE Transactions on Information and Systems
VolumeE99D
Issue number10
DOIs
Publication statusPublished - Oct 2016

Fingerprint

Recommender systems
Collaborative filtering
Learning systems
Feature extraction
Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Hardware and Architecture
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering
  • Artificial Intelligence

Cite this

Shilling attack detection in recommender systems via selecting patterns analysis. / Li, Wentao; Gao, Min; Li, Hua; Zeng, Jun; Xiong, Qingyu; Hirokawa, Sachio.

In: IEICE Transactions on Information and Systems, Vol. E99D, No. 10, 10.2016, p. 2600-2611.

Research output: Contribution to journalArticle

Li, Wentao ; Gao, Min ; Li, Hua ; Zeng, Jun ; Xiong, Qingyu ; Hirokawa, Sachio. / Shilling attack detection in recommender systems via selecting patterns analysis. In: IEICE Transactions on Information and Systems. 2016 ; Vol. E99D, No. 10. pp. 2600-2611.
@article{e9139465716042e59a2680397f93e7d1,
title = "Shilling attack detection in recommender systems via selecting patterns analysis",
abstract = "Collaborative filtering (CF) has been widely used in recommender systems to generate personalized recommendations. However, recommender systems using CF are vulnerable to shilling attacks, in which attackers inject fake profiles to manipulate recommendation results. Thus, shilling attacks pose a threat to the credibility of recommender systems. Previous studies mainly derive features from characteristics of item ratings in user profiles to detect attackers, but the methods suffer from low accuracy when attackers adopt new rating patterns. To overcome this drawback, we derive features from properties of item popularity in user profiles, which are determined by users' different selecting patterns. This feature extraction method is based on the prior knowledge that attackers select items to rate with man-made rules while normal users do this according to their inner preferences. Then, machine learning classification approaches are exploited to make use of these features to detect and remove attackers. Experiment results on the MovieLens dataset and Amazon review dataset show that our proposed method improves detection performance. In addition, the results justify the practical value of features derived from selecting patterns.",
author = "Wentao Li and Min Gao and Hua Li and Jun Zeng and Qingyu Xiong and Sachio Hirokawa",
year = "2016",
month = "10",
doi = "10.1587/transinf.2015EDP7500",
language = "English",
volume = "E99D",
pages = "2600--2611",
journal = "IEICE Transactions on Information and Systems",
issn = "0916-8532",
publisher = "一般社団法人電子情報通信学会",
number = "10",

}

TY - JOUR

T1 - Shilling attack detection in recommender systems via selecting patterns analysis

AU - Li, Wentao

AU - Gao, Min

AU - Li, Hua

AU - Zeng, Jun

AU - Xiong, Qingyu

AU - Hirokawa, Sachio

PY - 2016/10

Y1 - 2016/10

N2 - Collaborative filtering (CF) has been widely used in recommender systems to generate personalized recommendations. However, recommender systems using CF are vulnerable to shilling attacks, in which attackers inject fake profiles to manipulate recommendation results. Thus, shilling attacks pose a threat to the credibility of recommender systems. Previous studies mainly derive features from characteristics of item ratings in user profiles to detect attackers, but the methods suffer from low accuracy when attackers adopt new rating patterns. To overcome this drawback, we derive features from properties of item popularity in user profiles, which are determined by users' different selecting patterns. This feature extraction method is based on the prior knowledge that attackers select items to rate with man-made rules while normal users do this according to their inner preferences. Then, machine learning classification approaches are exploited to make use of these features to detect and remove attackers. Experiment results on the MovieLens dataset and Amazon review dataset show that our proposed method improves detection performance. In addition, the results justify the practical value of features derived from selecting patterns.

AB - Collaborative filtering (CF) has been widely used in recommender systems to generate personalized recommendations. However, recommender systems using CF are vulnerable to shilling attacks, in which attackers inject fake profiles to manipulate recommendation results. Thus, shilling attacks pose a threat to the credibility of recommender systems. Previous studies mainly derive features from characteristics of item ratings in user profiles to detect attackers, but the methods suffer from low accuracy when attackers adopt new rating patterns. To overcome this drawback, we derive features from properties of item popularity in user profiles, which are determined by users' different selecting patterns. This feature extraction method is based on the prior knowledge that attackers select items to rate with man-made rules while normal users do this according to their inner preferences. Then, machine learning classification approaches are exploited to make use of these features to detect and remove attackers. Experiment results on the MovieLens dataset and Amazon review dataset show that our proposed method improves detection performance. In addition, the results justify the practical value of features derived from selecting patterns.

UR - http://www.scopus.com/inward/record.url?scp=84989872797&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84989872797&partnerID=8YFLogxK

U2 - 10.1587/transinf.2015EDP7500

DO - 10.1587/transinf.2015EDP7500

M3 - Article

AN - SCOPUS:84989872797

VL - E99D

SP - 2600

EP - 2611

JO - IEICE Transactions on Information and Systems

JF - IEICE Transactions on Information and Systems

SN - 0916-8532

IS - 10

ER -