Influencing and aggregating agents ’ preferences over combinatorial domains

In a multi-agent context where a set of agents declares their preferences over a common set of candidates, it is often the case that such agents interact and exchange opinions before voting. In this initial phase, agents may influence each other and therefore modify their preferences, until hopefully they reach a stable state. Recent work has modelled the influence phenomenon in the case of voting over a single issue. Here we generalize this model to account for preferences over combinatorially structured domains including several issues. When agents express their preferences as CP-nets, we show how to model influence functions and to aggregate preferences by possibly interleaving voting and influence convergence.