How to choose a committee based on agents' preferences?

This is an accompanying paper for an invited presentation at the 4th International Conference on Behavioral, Economic, and Socio-Cultural Computing (BESC-2017): There are numerous situations where people (or, more broadly, agents) need to select a set of individuals based on preferences of these people (these agents). For example, democratic societies elect parliaments, judges in competitions choose finalists, companies choose their advisory boards. In this talk we argue that such settings can be modeled in the language of multiwinner elections. Specifically, in a multiwinner election we are given a set of candidates, a set of voters (with preferences over the candidates), and a target committee size. The goal is to choose a subset of candidates of a given size, in a way that is most satisfying for the voters. We show that exact meaning of the phrase “most satisfying” strongly depends on the context, but we argue that the language of committee scoring rules is sufficiently rich to capture many interesting interpretations of this phrase.