Incremental elicitation for social choice in multi attribute domains

Voting schemes are very interesting tools for preference aggregation and they can be applied to several contexts. In this work the attention is focused on group decision making and it is showed how it is possible to combine approximate winner determination and incremental elicitation in multi attribute domains. Dealing with incomplete voter preference can be very tricky but it has been already proved that the minimax regret solution concept can be effectively applied to robust approximation algorithms to propose winners; here we consider multi attribute candidates and we use a weight-based preference configuration to adapt and simplify the elicitation process. We describe in detail the winner selection mechanism and some possible incremental elicitation methods for a specific positional scoring rule: Borda Count. Furthermore the repeated application of one of the best strategy has been analyzed in order to efficiently obtain the complete ranking over the multiattribute alternatives in a system. Finally to prove the effectiveness of these methods, we show some practical results considering Pareto-optimal data sets