Abstract

In the past two decades researchers in autonomous agents and multi-agent systems, a research community in AI, have successfully used techniques from theoretical computer science to analyse and conceive methods for collective decision making. The field is now turning towards the design of social choice mechanisms such as participatory budgeting, where both the incentive structure and the computational properties need to be aligned for a successful and impactful process. After a rapid introduction to this research field, in this talk I will showcase the above mentioned synergy between computer science and economics by presenting recent work on designing measures to detect divisive issues from individual preferences. In a rank aggregation problem, members of a population rank issues to decide which are collectively preferred - examples range from voting for electing candidates to aggregating the results of search engine algorithms. We focus instead on identifying divisive issues that express disagreements among the preferences of individuals. We analyse the properties of these divisiveness measures and their relation to existing notions of polarisation. We also study their robustness under incomplete preferences and algorithms for the control and manipulation of divisiveness.

Reference

Umberto Grandi, “Measuring and controlling divisiveness in rank aggregation”, IAST General Seminar, Toulouse: IAST, May 23, 2023, 11:30–12:30, room Auditorium 4 (First floor - TSE Building).