• Trust Management and Reputation Systems
The Client : Javna agencija za raziskovalno dejavnost RS
Project type: Bilateral projects
Project duration: 2015 - 2017
  • Description
Trust is not essential ingredient only for prosperity of societies, but increasingly so for advanced information services and e-environments in general. In this latter domain, where it is referred to as computational trust, it has been researched now for roughly twenty years. The reasons are mainly the following two: E-solutions started to widely penetrate the Internet in the mid-nineties, so trust became a crucial issue in this emerging ubiquitous computing world, where "ordinary" signals and pieces of information (which are vital for trust formation processes) are no longer available. Next, research of this phenomenon required appropriate computing power and technologies, which became available to the required extent approximately at the same time. During these twenty years of computational trust management many methods emerged to model - to better or worse extent - the trust phenomenon. Among those that have received wider interest is also method called Qualitative Assessment Dynamics developed by FRI UL. Another important issue, directly related to trust, is reputation. Now that the notion of trust and related methodologies is getting clear, this does not hold true for reputation. How does this phenomenon emerge? How is it related to trust? Which are the feed-back loops that govern interdependence of trust and reputation? To find solutions in this area which is highly dynamic and non-linear, multiagent systems (MAS) are a promising tool, which enable research in this field from several perspectives, including belief revision, arguments based reasoning (e.g., modelling argument accrual with possibilistic uncertainty), and retransmission of information based on credibility. And these are the areas where Universidad Nacional del Sur is a strong partner. Therefore the proposed project collaboration will issue the following three core problems: -derivation of reputation on the basis of trust and feed-back processes between trust and reputation (i.e. beliefs); -research of reputation and trust dynamics in relation to decision processes; -real life applications (rules and policies for achieving coherent and trustfull societies, evaluation of honesty of opinions on e-forums). The methodologies that will be used: -use of methods developed by Argentinian partners (e.g., arguments based reasoning) and methods developed by Slovene partner (e.g., Qualitative Assessment Dynamics); -use of agents (MAS) based tools for simulations (like Jade and Alpha Agent developed by FRI UL). The time-line of research and expected outputs: -Year 1) clear derivation of concepts and system relationships between trust and reputation; -Year 2) definition of reference scenarios and experimental verification with emphasis on trust - reputation dynamics and its relation to decision making together with evaluation of honesty of opinions (e.g., on e-forums like e-Bay and Tripadvisor); -Year 3) the analysis of results and writing of a common research article that will be submitted to a quality SCI journal in a co-authorship between Argentinian and Slovene partners.