Generating unfair ratings to attack online rating systems

Abstract

Many online rating systems have been developed to provide its users a point of reference when

deciding which counter party they should choose to transact with. To encounter attacks to maliciously

manipulate those systems, many trust model researches have been in strong development. However

not a lot of efforts have been put into testing their performance: it is very difficult to compare their

performance against different types of attack. This project aims to provide a platform to do that.

In this project we developed a marketplace simulation software to allow user to test and compare the

robustness of their trust models[1]. The environment we built is similar to Amazon’s Marketplace in

which product prices are non-negotiable. The user will be able to configure the market place, plug in

ready-made trust models or develop their own using the platform’s facility and receive coherent data

about their performance.

During the project we strictly follow Test Driven software development methodology with various

carefully chosen technologies. The objectives, details of working process as well as the application’s

implementation will be discussed in this report.