DIRECT Project

In real world, trust is a relationship between two entities: it is one entity's belief on certain attributes about the other. More specifically, it's whether the subject entity is capable of performing some specific activities and whether the subject entity is able to deliver the result with an acceptable certainty.

So far the best known technology for building trust, reputation mechanisms are large-scale word-of-mouth communities in which individuals share opinions on a wide range of topics, including companies, products, services, and even world events. A reputation system gathers, distributes, and aggregates feedback about participants' behavior.

The goal of trust and reputation systems is to encourage trustworthiness in transactions by using past behavior as a publicly available predictor of likely future behavior.

DIRECT project aims to build a distributed reputation and trust management broker framework for E-Commerce and Ad-Hoc Network Applications, addressing the challenges of eliciting, evaluating and propagating reputation scattered in the community. The work is related to the referral network approach by using a network of brokers to propagate reputation information.

Distributed Broker

One of the major inherent complexity in the distributed trust and reputation management framework is the complicated reputation information eliciting, evaluating and propagating. Proved in by the real-world successful experience, a broker design which shields the application developer from low-level, tedious, and error-prone details can significantly reduce the software lifecycle costs. The broker developer can leverage the previous development expertise and capturing implementations of key patterns in reusable frameworks, rather than building them entirely from scratch for each use-case. Hence, in the framework, trust and reputation management brokers is adopted to alleviate the user's reputation management burden. By delegating trust management to brokers, individual users only need to ask their brokers about the reputation of a service before any transaction with a server. The only overhead for a user is the responsibility to share the reputation feedback with its broker.

The Trust Hierarchy

Reputation information in general is distributed within the community, and enough reputation samples need be collected in order to make an accurate prediction of the trustworthiness of one principal. In a distributed system, brokers cannot rely on its own resources to rate all servers when interacting with these servers. Collaboration among brokers is extremely important. Only through collaboration can the system identify untrustworthy servers promptly thus reducing the risk for everyone in the community. Consequently, the challenge of gathering enough information for making an informed judgment must be confronted in any distributed trust management framework design.

Motivated by this requirement, a novel hierarchical trust management framework is invented in DIRECT project to combine the strengths of distributed and centralized reputation management systems. It allows the users to gradually aggregate enough reputation samples according to specific requirements flexibly. The trust relationship is divided into three levels, the first level, the direct trust level, is for users that subscribe the same trust broker where there is a direct measurement of trust among all members. The second level, the connected trust level, is when two users utilizes two different trust brokers. So users must find information about each other through some distributed trust collection protocols. If there is not enough trust that can be gathered at this level, the third level, the institution trust level, relies on a centralized trust authority to provide global certified trust service about each other for decision making. If the third level still cannot meet the trust policy, users will have to use some trustless protocol to conduct business.

More information can be found in the Publications page.