Evaluating unfair ratings attacks in online rating systems

Today, the Internet has introduced and created many new features and applications that are very accessible, useful and helpful to internet users. While accessibility to the internet and its applications is largely justified across the globe in recent years with the discovery of wireless technology a...

Full description

Bibliographic Details
Main Author: Poh, Geok Khoon.
Other Authors: School of Computer Engineering
Format: Final Year Project (FYP)
Language:English
Published: 2013
Subjects:
Online Access:http://hdl.handle.net/10356/52300
Description
Summary:Today, the Internet has introduced and created many new features and applications that are very accessible, useful and helpful to internet users. While accessibility to the internet and its applications is largely justified across the globe in recent years with the discovery of wireless technology and improvement in broadband connection speed, there remains an intriguing question as to exactly how useful and helpful the internet is to its users; How true and accurate are the reviews on review websites if the reviewers are mostly anonymous users; Can we trust every rating made on a particular buyer or seller on E-Commerce Systems without knowing who they are in person. The report will first touch on a marketplace simulation program integrated and enhanced with a simple user interface (GUI), and it consists of the working algorithms of 6 trust models, 6 attack models and 3 performance metrics. Further explanation and detailed elaboration of the program will be covered in this section. The report then looks into the effectiveness and accuracy of 2 trust models, namely the TRAVOS (Trust and Reputation model for Agent-based Virtual OrganisationS) and WMA (Weighted Majority Algorithm) against a wide range of known attack models over 3 key performance metrics.