A Taxonomy to Express Open Challenges in Trust and Reputation Systems

Abstract

During the past decade, online trust and reputation systems have provided cogent answers to emerging challenges in the global computing infrastructures relating to computer and network security, electronic commerce, virtual enterprises, social networks and cloud computing. The goal of these systems in such global computing infrastructures is to allow entities to reason about the trustworthiness of other entities and to make autonomous decisions on the basis of trust. This requires the development of computational trust models that enable entities to reason about trust and to verify the properties of a particular interaction. The robustness of these mechanisms is one of the critical factors required for the success of this technology. In this paper, we briefly present characteristics of existing online trust and reputation models and systems through a multidimensional framework that can serve as a basis to understand the current state of the art in the area. The critical open challenges that limit the effectiveness of today's trust and reputation systems are discussed by providing a comprehensive literature review. Furthermore, we present a set of our contributions as a way to address some of these challenges.

Personalized Composition of Trustful Reputation Systems

“…Since most of the reputation systems proposed in literature use computation methods that are entirely built from scratch [2], well-established approaches are rarely considered and promising concepts get lost in the shuffle. In order to foster reusability of the particular components of reputation systems, Sänger and Pernul [3] propose a hierarchical component taxonomy of computation engines.…”

“…Since Resnick et al's paper [1] on the use of reputation systems to facilitate trust in Internet transactions, a vast number of reputation systems has been proposed. One major problem in this context is that most of them are completely designed from scratch and that established ideas are rarely considered [2]. To address this, Sänger and Pernul [3] decompose common reputation systems into single functional building blocks and describe and implement them in a publicly available repository.…”

Reusability for Trust and Reputation Systems

“…Thereto, we clarify the objectives of the identified classes (functions) and name common examples. Our analysis and selection of reputation systems is based on different surveys [8,13,6,2,1]. Figure 3.2 gives an overview of the primary and secondary classes identified.…”

“…by [2]) that most reputation-based trust models proposed in the academic community are built from scratch and do not rely on existing approaches. Only a few authors continue their research on the ideas of others.…”

“…Computation methods applied range from simple arithmetic over statistical approaches up to graph-based models involving multiple factors such as context information, propagation or personal preferences. A general problem is that most of the new introduced trust and reputation models use computation methods that are designed from scratch and rely on one novel idea which could lead to better solutions [2]. Only a few authors built on proposals of others.…”