Manipulation Robustness of Collaborative Filtering Systems

TL;DR

This paper analyzes the robustness of collaborative filtering algorithms against manipulation, showing that certain classes are more resistant and offering guidance for designing more secure systems.

Contribution

It introduces the concepts of linear and asymptotically linear algorithms as more manipulation-resistant, supported by theoretical and empirical evidence.

Findings

Nearest neighbor algorithms are highly susceptible to manipulation.

Linear and asymptotically linear algorithms show increased robustness.

Results guide future design of manipulation-resistant collaborative filtering systems.

Abstract

A collaborative filtering system recommends to users products that similar users like. Collaborative filtering systems influence purchase decisions, and hence have become targets of manipulation by unscrupulous vendors. We provide theoretical and empirical results demonstrating that while common nearest neighbor algorithms, which are widely used in commercial systems, can be highly susceptible to manipulation, two classes of collaborative filtering algorithms which we refer to as linear and asymptotically linear are relatively robust. These results provide guidance for the design of future collaborative filtering systems.