Collaborative case-based preference elicitation

Abstract

Preference elicitation is a well known bottleneck that prevents the acquisition of the utility function and consequently the set up of effective decisionsupport systems. In this paper we present a new approach to preference elicitation based on pairwise comparison. The exploitation of learning techniques allows to overcome the usual restrictions that prevent to scale up. Furthermore, we show how our approach can easily support a distributed process of preference elicitation combining both autonomy and coordination among different stakeholders. We argue that a collaborative approach to preference elicitation can be effective in dealing with non homogeneous data representations. The presentation of the model is followed by an empirical evaluation on a real world settings. We consider a case study on environmental risk assessment to test with real users the properties of our model. © Springer-Verlag Berlin Heidelberg 2005.