Exploring Trust in Personal Learning Environments Na Li, Maryam Najafian-Razavi, Denis Gillet Ecole Polytechnique Fédérale de Lausanne (EPFL) 1015 Lausanne, Switzerland {na.li, maryam.najafian-razavi, denis.gillet}@epfl.ch Abstract—The design of effective trust and reputation mechanisms for personal learning environments (PLEs) is believed to be a promising research direction. In this paper, we propose a 4-dimensional trust model that complies with the specific requirements of PLEs. Trust is explored in four dimensions: trustor, trustee, context and visibility. The importance of these four dimensions is investigated through a number of scenarios. The model is implemented in a PLE platform named Graaasp. Preliminary evaluation of usefulness is conducted through a user study and some interesting findings are discussed in the end. Keywords-trust; reputation; personal learning environment; rating; ranking I. INTRODUCTION Benefiting from the success of Web 2.0 social media, interactive information sharing has become pervasive. For users surrounded by an abundance of information, the challenge now is how to determine which resources can be relied upon and who is reliable enough to interact with. To solve this problem, a number of trust and reputation systems have been developed in various platforms, including e- commercial sites, product review systems, and professional communities. Trust and reputation measures can help users decide whether or not to interact with a given party in the future, or whether it is safe to depend on a given resource [1]. This creates an incentive for good behavior, therefore inducing a positive effect on the quality of interaction in online communities. As a particular support framework for interaction in online communities, personal learning environments (PLEs) embed tools, services, content and people involved in the digital part of the learning process [2] [3]. Web 2.0 functionalities like blogging, tagging, rating and commenting are gradually incorporated into learners’ overall learning ecology, contributing to increasing learning incentives and enhancing the learning experience [4]. On one hand, these Web 2.0 features enable learners to express opinions easily and facilitate accumulating domain knowledge. On the other hand, learners’ active contributions produce a large amount of user-generated content, which may lead to information overflow. In such an open learning environment, it is not easy for learners to find suitable people to learn from or collaborate with. Moreover, the flood of data might bring about the challenge of selecting useful learning resources depending on personal learning goals. Therefore, research efforts are needed to design appropriate trust and reputation mechanisms for PLEs, aiming at expertise assessment and quality assurance to support self-directed learning activities. As a complex social concept, trust reflects the subjective perception one party holds about another party. It’s asymmetrical, transferable, dynamic, and context-dependent, and can be influenced by various factors. Trust has been interpreted in different ways to comply with specific requirements of various domains. In this paper, a usable trust model for personal learning environments is investigated and implemented in a PLE prototype named Graaap 1 . Preliminary evaluation of usefulness has been conducted through a user study. The rest of this paper is organized as follows. Section II introduces the existing trust and reputation schemes, and discusses the specific aspects of PLEs that call for different trust and reputation models. A trust model dedicated to PLEs is proposed in Section III. Section IV introduces the Graaasp prototype and illustrates the implementation of the proposed trust model in it. User evaluation and main findings are addressed afterwards in Section V. Finally, Section VI concludes the paper and discusses the future work. II. RELATED WORK In order to develop effective trust and reputation mechanisms, a number of attempts have been made in both literature and practice. The simplest but most widely used scheme is to compute an average or summary of all ratings for an entity. The reputation systems used by eBay 2 , Epinions 3 and Amazon 4 fall into this category. However, this scheme is primitive and therefore gives a poor picture on an entity’s reputation score [5]. Google’s PageRank [6], Advogato’s reputation scheme [7], and EigenTrust model [8] can be categorized as flow models, where trust or reputation is computed by transitive iteration through looped or arbitrarily long chains. In short, a participant’s trust or reputation score increases as a function of incoming flow, and decreases as a function of outgoing flow. Flow model-based schemes adopt global trust metrics, where a single trust or reputation score is associated with each participant and displayed to all members in the community. Some researchers have proposed to use mathematical models in an attempt to measure trust, including Bayesian algorithm-based metrics [9] and belief theory-based models 1 Graaasp (graaasp.epfl.ch): a PLE prototype. 2 eBay (www.ebay.com): an online auction and shopping website. 3 Epinions (www.epinions.com): a consumer review website 4 Amazon (www.amazon.com): an e-commercial website. 274 ACHI 2011 : The Fourth International Conference on Advances in Computer-Human Interactions Copyright (c) IARIA, 2011. ISBN: 978-1-61208-117-5