2017/07/07 by Manuele Leonelli, Leonelli, Manuele, Eva Riccomagno +3 · 1 citation
Computer Science · Decision Sciences · #Bayesian Modeling and Causal Inference #Cognitive Science and Mapping #FOS: Computer and information sciences #Other Statistics (stat.OT) #Scientific Computing and Data Management
paper · pdf · doi:10.48550/arxiv.1707.02070
openalex publication_date 2017/07/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Current decision support systems address domains that are heterogeneous in\nnature and becoming progressively larger. Such systems often require the input\nof expert judgement about a variety of different fields and an intensive\ncomputational power to produce the scores necessary to rank the available\npolicies. Recently, integrating decision support systems have been introduced\nto enable a formal Bayesian multi-agent decision analysis to be distributed and\nconsequently efficient. In such systems, where different panels of experts\noversee disjoint but correlated vectors of variables, each expert group needs\nto deliver only certain summaries of the variables under their jurisdiction to\nproperly derive an overall score for the available policies. Here we present an\nalgebraic approach that makes this methodology feasible for a wide range of\nmodelling contexts and that enables us to identify the summaries needed for\nsuch a combination of judgements. We are also able to demonstrate that\ncoherence, in a sense we formalize here, is still guaranteed when panels only\nshare a partial specification of their model with other panel members. We\nillustrate this algebraic approach by applying it to a specific class of\nBayesian networks and demonstrate how we can use it to derive closed form\nformulae for the computations of the joint moments of variables that determine\nthe score of different policies.\n