An algorithm for detecting recommendation attack based on incremental learning

作者：Yu Hongtao^*; Li Peng; Zhang Fuzhi

来源：Journal of Information and Computational Science, 2014, 11(7): 2365-2373.

DOI：10.12733/jics20103402

• Attack detection
• Attributes
• Detection models
• Incremental learning
• Recommendation attacks
• Statistical features
• Training sets
• User profile

The existing supervised approaches suffer from the problem that detection model cannot update incrementally with the increase of user profiles. Aiming at this problem, we propose a recommendation attack detection algorithm based on incremental learning by introducing the rough set theory. Firstly, an algorithm for building a training set is proposed to choose the best label samples used to build a classifier. Secondly, newly labeled samples are used to train the classifier incrementally in order to make it more reasonable to cover attack profiles. Finally, the attack detection method based on statistical features is proposed to distinguish attack profiles from genuine user profiles. The experimental results on the MovieLens dataset show that the proposed algorithm can effectively improve the performance of attack detection.

• 出版日期2014