State of the art of Trust and Reputation Systems in E-Commerce Context

Abstract-This article proposes in depth comparative study of the most popular, used and analyzed Trust and Reputation System (TRS) according to the trust and reputation literature and in terms of specific trustworthiness criteria. This survey is realized relying on a selection of trustworthiness criteria that analyze and evaluate the maturity and effectiveness of TRS. These criteria describe the utility, the usability, the performance and the effectiveness of the TRS. We also provide a summary table of the compared TRS within a detailed and granular selection of trust and reputation aspects.

I. INTRODUCTION Open electronic markets, online collaboration systems, distributed peer-to-peer applications, online social media require the establishment of mutual trust between service providers and service consumers. In fact, the major concerns of web-based services especially e-commerce applications is to overcome the inherent uncertainties and untrustworthiness risks and enhance the system’s robustness and resistance against fraudulent users and distrustful ones. Besides, e-commerce platforms aim at adopting the most efficient approach that helps detect and analyze users’ intentions in order to reveal and understand deceitful ones. Otherwise, the underlying purpose of e-commerce services which is to maximize the profit and the rate of purchase, would be threatened and deteriorated by fraudulent and ill-intentioned users. For this reason, Recommender Systems such as Trust and Reputation Systems (TRS), provide essential input for computational trust so as to predict future behaviors of peers basing on the past actions of a peer [1]. In a reputation network, information about these actions can also be received from other members of a reputation network who have transacted with the peer. However, the credibility of this third-party information must be critically assessed. The underlying goal in all reputation systems is to predict a peer’s future transactions taking into account his past actions and applying algorithms relying on probabilities approaches.
To gather these first-hand transactions is a tedious and costly task especially when involving the malicious and fraudulent peers. To overcome this limitation, users share their experiences through the reputation system, which aims to detect and effectively isolate misbehaving customers and users. Indeed, e-commerce users refer to this shared information as recommendations on which they rely in order to make the right purchase decision. As no user in e-commerce environment is fully trusted, recommendations and reviews credibility and trustworthiness must be critically assessed. At this purpose, Trust and Reputation Systems have been widely used for various e- commerce applications in order to assess the credibility and trustworthiness of the provided reputation information. Furthermore, this task is becoming increasingly important for the majority of e-services, but especially for e-commerce platforms where resources and business benefice value depend on making correct decisions. As a matter of fact, deliberately providing fake and dishonest ratings and reviews is a serious and crucial security issue that threaten the trust establishment and propagation in e-commerce environment. In fact, this misbehaving attitude would systematically falsify the trust and reputation assessment of the reviewed products and services in e-commerce applications. As a result, this falsification would impact customers’ trust with regards to the purchase decision-making. Moreover, human users have specific reasons for deliberately skewing their comments and they can change their intentional behavior over time and according to changing circumstances that impact the product’s quality and the customer’s interests as well. Besides, customers can also be discriminative against particular service providers while cooperating with others. In order to distinguish honest reputation information from dishonest one, we need a robust TRS that applies intelligent detection algorithms, either supervised, unsupervised or semi-supervised. These algorithms aim to analyze the trustworthiness of the reputation information provided in the form of reviews, recommendations and numeric ratings. In fact, a meticulous trustworthiness analysis of the shared information achieved by TRS, would certainly increase the system’s robustness and resistance against fraudulent reviewers. Moreover, the underlying computational goal of TRS is to generate a trustful evaluation of the reputation of a product or a service in e- commerce. Indeed, TRS incarnate a combination of two dependant systems: Reputation systems which is generally ba

…(Full text truncated)…