A Bayesian Model for Event-based Trust

Trust models in ubiquitous computing

“…to formalize this preference towards newer information, it is not clear what the exact benefits of forgetting factors are, and more generally when and why to choose between, for example, exponential decay as opposed to say linear decay? Nielsen et al (2007) propose preliminary ideas following the spirit of this paper are introduced, formally modelling the dynamic behaviour of a principal by a hidden Markov model.…”

“…and (ii) to express and argue how robust a particular system is with respect to changes in the environment. In this section, we illustrate a few preliminary attempts towards such a framework based on ideas from , Nielsen et al (2007) and Sassone et al (2007).…”

“…in estimating the fixed probability q of honest behaviour of a particular principal. Nielsen et al (2007) also illustrate how the approach addresses question (ii) above. The two algorithms are generalized to a continuum of algorithms, and it is shown that among this continuum of algorithms, the beta-based algorithm of Mui et al (2002) is optimal for precisely qZ 1=2G1= ffiffiffiffiffi 12 p .…”

“…This framework has been applied by Nielsen et al (2007) to ask and to answer formally, for example, the question of comparing the performances of the maximum-likelihood algorithm of Despotovic & Aberer (2004) and the betabased algorithm of Mui et al (2002) relative to the simple l model introduced above, i.e. in estimating the fixed probability q of honest behaviour of a particular principal.…”

Events, Causality and Symmetry

“…where numerous researchers have used or rediscovered causal models in providing a compositional semantics-see [76] and its references; • Network diagnostics: in the monitoring and fault diagnosis of communication networks [10]; • Logic of programs: in concurrent separation logic where some artificialities in Brookes' pioneering soundness proof are obviated through a Petri-net model [49]; • Partial order model checking: following the seminal work of McMillan [64] the unfolding of nets is exploited in the automated analysis of systems [32]; • Distributed computation: event structures appear both classically [60] and recently in the Bayesian analysis of trust [71].…”