Collaborative Shilling Detection Bridging Factorization and User Embedding

Dou, Tong; Yu, Junliang; Xiong, Qingyu; Gao, Min; Song, Yuqi; Fang, Qianqi

doi:10.1007/978-3-030-00916-8_43

Cited by 16 publications

(14 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where , W (1) , and p denote the activation function, the weight matrix, and the bias vector, respectively. In a similar way, the AE tries to reconstruct the input vector from the hidden representation by the following formula:x = (W (2)…”

Section: Sparse Denoisingmentioning

confidence: 99%

“…wherex, , W (2) , and q denote the vector of decoding layer, the activation function, the weight matrix, and the bias vector, respectively.…”

Section: Sparse Denoisingmentioning

confidence: 99%

“…where = {W (1) , W (2) , p, q} and () represents the loss function. A sparse autoencoder (SAE) is an AE that requires that most of the neurons are zero.…”

Section: Sparse Denoisingmentioning

confidence: 99%

“…In the methods based on multiple views, from the information of rating values and corated items, Dou et al [2] proposed a collaborative shilling detection model, which learned the latent factors by decomposing the user-item rating matrix and the user-user cooccurrence matrix. After that, the latent factors were used as features, and decision tree was used as a classifier algorithm to detect attackers.…”

Section: Related Workmentioning

confidence: 99%

“…Generally, detection of shilling attacks can be seen as a binary classification problem; that is, for each user profile, the classified result can be a normal user or an attacker [2]. Therefore, a number of detection features have been proposed and some machine learning methods are used to separate the attackers from genuine users.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Detecting Shilling Attacks with Automatic Features from Multiple Views

Hao

Zhang

Wang

et al. 2019

Security and Communication Networks

View full text Add to dashboard Cite

Due to the openness of the recommender systems, the attackers are likely to inject a large number of fake profiles to bias the prediction of such systems. The traditional detection methods mainly rely on the artificial features, which are often extracted from one kind of user-generated information. In these methods, fine-grained interactions between users and items cannot be captured comprehensively, leading to the degradation of detection accuracy under various types of attacks. In this paper, we propose an ensemble detection method based on the automatic features extracted from multiple views. Firstly, to collaboratively discover the shilling profiles, the users’ behaviors are analyzed from multiple views including ratings, item popularity, and user-user graph. Secondly, based on the data preprocessed from multiple views, the stacked denoising autoencoders are used to automatically extract user features with different corruption rates. Moreover, the features extracted from multiple views are effectively combined based on principal component analysis. Finally, according to the features extracted with different corruption rates, the weak classifiers are generated and then integrated to detect attacks. The experimental results on the MovieLens, Netflix, and Amazon datasets indicate that the proposed method can effectively detect various attacks.

show abstract

Section: Sparse Denoisingmentioning

confidence: 99%

“…wherex, , W (2) , and q denote the vector of decoding layer, the activation function, the weight matrix, and the bias vector, respectively.…”

Section: Sparse Denoisingmentioning

confidence: 99%

“…where = {W (1) , W (2) , p, q} and () represents the loss function. A sparse autoencoder (SAE) is an AE that requires that most of the neurons are zero.…”

Section: Sparse Denoisingmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations