Instilling similarity rating model to combat malicious raters

The rapidly growing popularity of auction and rating websites has given rise to an increase use of ratings in online environment. However, many of these websites do not take into consideration several issues in ratings such as malicious raters, different user perceptions, number of available ratings and others. Thus, this creates mistrust between online buyers and products/services that the websites offer. In this paper, we discuss various issues faced by the current online rating systems. We then propose a rating model, termed as CPR (Credible and Personalised Rating), to address some of these issues. We demonstrate how our rating model can be implemented in both centralized and decentralized architectures. In addition, several experiments are conducted to study the effectiveness of our rating model to combat malicious and bogus ratings.