[bull-ia] PhD proposal: Elicitation and explanation for voting rules

PhD proposal: Elicitation and explanation for voting rules

A competition for a full time three-year long Ph.D.
position in computer science at LAMSADE, Université
Paris-Dauphine is open.

This call is also available as a PDF.

Keywords: artificial intelligence,
computational social choice, decision theory, axiomatic
analysis, preference elicitation, preference modeling,
formal argumentation theory

Context

The theme of this call is around the development of
procedures to help a committee (or a society) choose a
suitable voting rule. This will involve 1) axiomatic
analysis of voting rules; 2) explanation of those axioms in
terms a non-expert can understand; 3) preference elicitation
methods. As such, the dissertation will be situated at the
interface between computer science and economics.

Voting rules are formal means to aggregate preferences of a
group of voters into a collective decision. Social choice
theory has produced and analyzed several voting rules, many
of which being apparently reasonable: no rule can be
considered “the best voting rule”, independent of the
context. However, some rules are, upon reflexion, better
than others. As a result, the choice of the voting rule to
be used turns out to be an interesting question. For
example, in a (social or commercial) enterprise a body may
have to decide how conflicts will be sorted out, or a
parliament may decide about a possible revision of the
electoral law of the country.

More details about the related research context are
available in the long
version of this call.

Expected
results and impact

The PhD will strive to develop explainable models and
justifiable recommendations. It may thus contribute to make
scientific-driven recommendations more acceptable to the
general public. The results are expected to be published in
top-ranked conferences and journals in the AI and social
choice fields.

Besides theoretical work, it might aim at producing
practical methods that could be integrated into the Whale
platform. As an example, automatic deduction might be used
to automatically illustrate various (paradoxical or
intuitive) results when applying known voting rules to
concrete situations.

Expected
assets

The subject involves social choice theory, decision theory,
formal argumentation theory. The candidate must have strong
competencies in some of these fields, and be interested in
the other ones. Regarding language requirements, a good
command of English and willingness to learn French in
necessary.

Environment
and contacts

Applicants must send their CV, coordinates of a referent
person, a letter of motivation and a listing of the courses
they took in masters, together with their marks, to remzi.sanver@dauphine.fr
until 1 April 2018. The position is to
start on 1 October 2018.

The successful student will be hosted at LAMSADE,
Université Paris-Dauphine (which is a part of Université
Paris Sciences et Lettres) and will be supervised by Remzi
Sanver and Olivier Cailloux.