Study On Standard Reputation Model For C2C Online Trading System

With the popularity of the Internet and the rapid growth of the Internet users, online trading has been greatly developed. However, opportunistic behavior led to many frauds, the lack of credibility has become an important factor that blocks the online trading. Online reputation system is a new trust mechanism, whose aim is to establish, maintain the trust relationship, and protect the online market order.The reputation model is the core of the reputation system. It provides the reputation system a mean to measure the rating, aggregates the ratings from reputation system in a certain rules and, updates the reputation value of the target user. It's the fatal factor that impacts the effects of the online reputation system.This paper analyzes the popular reputation models, then proposes a average-aggregated-based reputation model, named Standard Reputation Model and, studies the impact factors of the Standard Reputation Model; finally simulates the availability of it.First of all, this paper analyzes the popular reputation models in three aspects: the rating measurements method, the rating aggregation method, concludes that a simple rank rating method can not meet the online trading requirement for the reputation system; the rating method requires greater ability to distinguish the reputation. The aggregation methods of reputation include the add-up-aggregated method and average-aggregated method, the former is widely used, while the later is rarely used for its lack of ability to distinguish the various reputation situation and sensitive nature to the changes of reputation.Based on the study of popular reputation models, a model, using the average-aggregated-based method, named Standard Reputation Model, is proposed. It uses a real number belong to [0, 1], thus it has an unlimited ability to distinguish the various reputation situation which is the most important weakness of the average-aggregated model. Compared with the add-up-aggregated-based reputation model including negative rating, the Standard Reputation Model is more acceptable for its nonnegative rating. This paper also analyzes the law and gives a setting formula about the importance of the past ratings. The paper gives how to computing the credibility of the rating, too.Finally, I develop a simulation system with Matlab 7 to compare the proposed model with the eBay reputation model. The experimentations show that the Standard Reputation Model is more efficient against malicious behaviors; it also proves that average-aggregated-based reputation model is unstableness via setting the factors properly.The proposed model unifies the reputation of all users to the same range, which makes reputations of all users are comparable, so that the online reputation system can pay its role better, so as to promote the development of online trading.