Arguing about informant credibility in open multi-agent systems

Abstract

This paper proposes the use of an argumentation framework with recursive attacks to address a trust model in a collaborative open multi-agent system. Our approach is focused on scenarios where agents share information about the credibility (informational trust) they have assigned to their peers. We will represent informants' credibility through credibility objects which will include not only trust information but also the informant source. This leads to a recursive setting where the reliability of certain credibility information depends on the credibility of other pieces of information that should be subject to the same analysis. Credibility objects are maintained in a credibility base which can have information in conflict. In this scenario, we will formally show that our proposal will produce a partially ordered credibility relation; such relation contains the information that can be justified by an argumentation process.Example 1 (Running example). Let Alex, Barbara, and Carla (abbreviated A, B, C) be stock market experts, and let Harry, John, and Kate (abbreviated H, J, K) be journalists on this topic. Agent Tory (T ) has all the agents mentioned above as its informants, and T has collected the following information about their credibilities in the stock market topic. According to Kate, Alex is more credible than Carla (represented A > C); and according to Harry, Carla is more credible than Barbara, (C > B). However, for John, Barbara is more credible than Alex, (B > A). Finally, according to the editor of a newsletter called X-market, which specializes on this topic, the journalist Kate is more credible than John, (K > J). Now, assume that Tory has to decide whether Alex is more credible than Barbara. Tory can conclude B > A using the information obtained from John but, assuming the credibility relation is transitive, Tory can also infer A > B from the information obtained from Kate (A > C) and Harry (C > B). In this situation, Tory can use the information that she has about the credibility of the journalists John, Harry and Kate to decide whether A > B or B > A prevails. Since the editor of X-market reports that K > J, then Tory has a reason to consider Kate more credible than John on this topic and therefore, a reason to support that Alex is more credible than Barbara.

A Survey of Adaptive Multi-Agent Networks and Their Applications in Smart Cities

“…For example, in swarm optimization, agents are categorized and distinguished based on the quality of their solutions in the previous iteration [29]. In the IoTs systems, some agents and elements may be selfinterested, with a lack of a global perspective of the system [30], and the potential to inject unreliable and misleading information into the system, which needs extra considerations during modeling and evaluation states [31].…”

Acquiring knowledge from expert agents in a structured argumentation setting

“…The consistency question from [46] together with the proposal of [25,39,40] could be used in our formalism to add an implementation for the operator reject-expert-opinion defined in Section 5. The authors consider that agents can obtain information from multiple informants, and that the attribution of trust to a particular informant can be higher than the trust attributed to others.…”

Optimizing group learning: An evolutionary computing approach

“…Thus, the question to be investigated is how members of a research group should update on the receipt of new evidence in a social setting, where they also have access to relevant beliefs of (some or all of) their colleagues, supposing the group wants to strike the best balance between speed (getting at the truth fast) and accuracy (minimizing error rates). The main methodological tool to be used is that of computational agent-based modelling, which has become a central topic in artificial intelligence (Shoham, Powers, & Grenager, 2007;Tamargo, Garcia, Falappa, & Simari, 2014;Nunes & Antunes, 2015;Gottifredi et al, 2018). Specifically, we build on the well-known Hegselmann-Krause (HK) model for studying opinion dynamics in groups of interacting agents focused on a common research question (Hegselmann & Krause, 2002, 2005, 2009; for related models, see Deffuant et al, 2000;Dittmer, 2001;Weisbuch et al, 2002;Pluchino, Latora, & Rapisarda, 2006;Semeshenko, Gordon, & Nadal, 2008;De Langhe & Greiff, 2010).…”

“…In this model, and also in Douven and Wenmackers' extension of it to be discussed below, all agents in the BCI are treated on a par. If information about the credibility of these agents were available, one might plausibly wish to weigh them differently, for instance, in the manner ofTamargo et al (2014) andGottifredi et al (2018).…”