Reputation model based on source credibility theory and its applications

In this brief material, we introduce some works on source credibility theory, including trust model, reputation evaluation, rater system and applications.

A Reputation-Based Trust Model for Peer-to-Peer eCommerce Communities”[1]. Peer-to-Peer (P2P or C2C) eCommerce communities are commonly perceived as an environment offering both opportunities and threats. One way to minimize threats in such an open community is to use community-based reputations. The authors present a coherent adaptive trust model for quantifying and comparing the trustworthiness of peers based on a transaction-based feedback system. There are two main features of the model. First, the trust models based solely on feedback from other peers in the community is inaccurate and ineffective. They introduce three basic trust parameters in computing trustworthiness of peers:  1) feedback information; 2) the total number of transactions; and 3) the credibility of the feedback sources. Second, they introduce two adaptive factors: 1) the transaction context factor and 2) the community context factor.

A survey of trust and reputation systems for online service provision” [2]. A simple review paper, it gives an overview of existing and proposed systems that can be used to derive measures of trust and reputation for Internet transactions, to analyse the current trends and developments in this area, and to propose a research agenda for trust and reputation systems.

 

A Reputation Evaluation Technique for Web Services”[3], in this paper, a trust and reputation management framework for web service selection is proposed. In which, reputation evaluation algorithm is based on similarity theory. Similarities and trusts are used as weights for computing reputations from different recommenders.

A good trust and reputation framework should consider following issues simultaneously: 1) reputations for a new added service should be valuated. 2) different weights should be assigned to reputations from different recommenders. 3) trusts and reputations should be updated after each interaction.

 

A Reputation Mechanism for Business-to-Business Electronic Commerce That Accounts for Rater Credibility”.[4] In contrast to consumer electronic marketplaces, raters in B2B communities are skilled and connected, necessitating a reputation mechanism to account for the relationship between the user and the rater. So, in this paper, the authors presented a prototype rating tool: TrustBuilder which incorporate a methodology to calculate a weighted rating based on source credibility theory. This solution offers several advantages over existing models. 1) source credibility theory provides tested frameworks for aggregating ratings from different sources. 2)  there are validated scales for measuring a source’s (rater’s) credibility. 3) weights of a rater’s ratings depend on user preferences instead of rater behavior.

According to the source credibility theory, the credibility of an information source comprises expertise (competency), trustworthiness, co-orientation (similarity), and attraction. (four key factors, attraction is not appropriate for the online reputation system because it is generated in the environment where an information source is revealed to an information receiver, first three factors are considered)

 

Q-rater: A collaborative reputation system based on source credibility theory”[5]. Unlike consumer-to-consumer (C2C) e-commerce sites, most business-to-consumer (B2C) sites do not provide users with explicit information on the reputation of a user as assessed by other users. The possibility of incorrect information from unreliable users and declining trust in the electronic market looms large. Hence, there is a need for a fairer and more objective rating mechanism for a rater. The authors propose an online reputation system suitable for B2C e-commerce sites where both explicit user evaluation ratings and social relationships among users are inadequate. The conceptual framework of their proposed mechanism is based on the source credibility theory for WOM (Word-Of-Mouth) communications model in consumer psychology.

User reputation generating mechanisms can be classified into two different types depending on the e-business model: 1) bidirectional rating mechanism for enterprises with the C2C e-business model (e.g. eBay and Napster). 2) unidirectional rating mechanism for enterprises with the B2C e-business model (e.g. Amazon.com, Epinion.com)

In [4], users had to assign explicit ratings directly to other users. They focused only on generating the reputation of the raters. However, in Q-rater system[5], the credibility of users is implicitly extracted from their past rating data in order to conduct a more objective calculation of their reputations. Q-rater focuses on generating both user and item reputations implicitly.

The overall process of Q-Rater is divided into three phases: ‘‘user credibility extraction phase,” ‘‘user reputation generation phase” and “item reputation generation phase.” In the first phase, the source credibility factor of each user is measured implicitly from the user’s explicit ratings for the items. In the second phase, the reputation of each user is generated by a combination of three source credibility factors and the qualified rater group is formed by the user reputation. The item reputation generation phase is executed by the rating tendency calculation and the item rating aggregation to predict general users’ evaluation for each item. Through forming the qualified rater group, the item’s reputation will be generated based on the reputation-weighted rating aggregation mechanism.

 

A probabilistic reputation model based on transaction ratings” [6]. This work introduces a probabilistic model allowing to compute reputation scores as close as possible to their intrinsic value. It is based on consumer–provider interaction model. In which, the providers supply the items with a quality following a normal law, centered on their intrinsic ‘‘quality of service”. The consumers, after the reception and the inspection of the item, rate it according to a linear function of its quality – a standard regression model. Two extensions of this basic model are considered as well: a model accounting for truncation of the ratings and a Bayesian model assuming a prior distribution on the parameters. The experiments suggest that these models are able to extract useful information from the ratings, are robust towards adverse behaviors such as cheating, and are competitive in comparison with standard methods.

 

Modeling Decentralized Reputation-Based Trust for Initial Transactions in Digital Environments” [7]. Evaluating trustworthiness of a service provider without any prior historical transactions (i.e. the initial transaction) poses a number of challenging issues. This article presents TIDE – a decentralized reputation trust mechanism that determines the initial trustworthiness of entities in digital environments. TIDE improves the precision of trust computation by considering raters’ feedback, number of transactions, credibility, incentive to encourage raters’ participation, strategy for updating raters’ category, and safeguards against dynamic personalities. Furthermore, TIDE classifies raters into three categories and promotes the flexibility and customization through its parameters. Evaluation of TIDE against several attack vectors demonstrates its accuracy, robustness and resilience.

 

A Case Study of Collaboration and Reputation in Social Web Search” [8]. In this paper, they propose a reputation model for HeyStaks[9] users that utilize the implicit collaboration events that take place between users as recommendations are made and selected. They describe a live-user trial of HeyStaks that demonstrates the relevance of its core recommendations and the ability of the reputation model to further improve recommendation quality. Their findings indicate that incorporating reputation into the recommendation process further improves the relevance of HeyStaks recommendations by up to 40%.

 

A comparative study of collaboration-based reputation models for social recommender systems”[10]. In this paper, the authors describe a generic approach to modeling user and item reputation in social recommender systems. In particular, they show how the various interactions between producers and consumers of content can be used to create so-called collaboration graphs, from which the reputation of users and items can be derived. They analyze the performance of their reputation models in the context of the HeyStaks social search platform, which is designed to complement mainstream search engines by recommending relevant pages to users based on the past experiences of search communities.

 

HeyStaks [9] is a company that enables people to implicitly collaborate and share in each other's search & browsing experiences. By automatically forming communities of users who share similar contexts, a powerful form of recommendation is possible and deep insights into the user base can be delivered, such as key influencers/experts, gaps in information, etc. The company is a startup based in Dublin and was founded by Barry Smyth, Peter Briggs and Maurice Coyle.

 

References

1.             Xiong, L. and L. Liu, A Reputation-Based Trust Model for Peer-to-Peer eCommerce Communities, in Proceedings of the 4th ACM conference on Electronic commerce 2003, ACM, New York. p. 228-229

2.             Jøsang, A., R. Ismail, and C. Boy, A survey of trust and reputation systems for online service provision. Decision Support Systems, 2007. 43: p. 618–644.

3.             Yang, N., X. Chen, and H. Yu, A Reputation Evaluation Technique for Web Services. International Journal of Security and Its Applications, 2012. 6(2): p. 329-334.

4.             Ekstrom, M.A., H.C. Bjornsson, and C.I. Nass, A Reputation Mechanism for Business-to-Business Electronic Commerce That Accounts for Rater Credibility. Journal of Organizational Computing and Electronic Commerce, 2005. 15(1): p. 1-18.

5.             Cho, J., K. Kwon, and Y. Park, Q-rater: A collaborative reputation system based on source credibility theory. Expert Systems with Applications, 2009. 36: p. 3751–3760.

6.             Fouss, F., Y. Achbany, and M. Saerens, A probabilistic reputation model based on transaction ratings. Information Sciences, 2010. 180: p. 2095–2123.

7.             PRANATA, I., R. ATHAUDA, and G. SKINNER, Modeling Decentralized Reputation-Based Trust for Initial Transactions in Digital Environments. ACM Transactions on Internet Technology, 2013. 12(3): p. Article 8.

8.             MCNALLY, K., et al., A Case Study of Collaboration and Reputation in Social Web Search. ACM Trans. Intell. Syst. Technol, 2011. 3(1): p. Article 4.

9.             http://www.heystaks.com/.

10.          McNally, K., M.P. O'Mahony, and B. Smy, A comparative study of collaboration-based reputation models for social recommender systems. User Model User-Adap Inter, 2013.