In these recent years my research activities evolved around two main subjects:
The idea here is to combine research and findings from Decision Analysis and Artificial Intelligence in an attempt to be able to handle new challenges in Decision Support (presence of very large data sets, streaming information, combinatorial structures of alternatives, uncertainty management, qualitative decision making etc.) as well as in Artificial Intelligence such as: automatic decision devices, reasoning in presence of incomplete and/or ambiguous information, argumentation etc..
<![if !supportLists]>- <![endif]>What is a decision problem? We try to establish a general framework within which the definition of what is a decision problem should be independent from the methods used in Operational Research, Decision Analysis, Artificial Intelligence etc.. This work is conducted with Alberto Colorni and you can see our papers in ADT 2013 and in 2018. Within this general framework I am specifically interested to the problem of generating alternatives before these being evaluated.
<![if !supportLists]>- <![endif]>Non-conventional preference modelling: Preference models under uncertainty and/or ambiguity. Axiomatization of partial comparability. Axiomatization of preference structures admitting a threshold representation. Generalisation of the concordance - discordance principle in preference aggregation. Positive and Negative reasons in decision aiding. Hierarchical preference aggregation. Exploitation of non-conventional preference structures. Applications in negotiation support.
<![if !supportLists]>- <![endif]>Convincing Rating. Most of the existing rating procedures can result to counterintuitive results due to the typical paradoxes of ordinal preference aggregation. In recent work we propose new methods which produce “convincing ratings” merging supervised learning methods and the explicit use of positive and negative reasons when ratings need to be computed.
<![if !supportLists]>- <![endif]>Algorithmic Fairness. The issue of “Social Responsibility of Algorithms” (www.lamsade.dauphine.fr/sra2017) is becoming increasingly important both for scientists and the society. Among the topics which draw attention is “fairness in algorithms”. Our approach is to study this topic under a social choice theory point of view where fairness has been extensively studies and whether it is possible to consider the design of algorithms under the perspective of mechanism design.
<![if !supportLists]>2. <![endif]>Policy Analytics
I introduced this concept some years ago and make it more precise in the position paper we published to the inaugural issue of the EURO Journal on Decision Processes (see in the papers). The idea here is to develop a comprehensive "Decision Aiding Methodology" (combining theoretical findings, empirical evidence and best practices) with particular emphasis to the problem of conceiving, analysing, assessing, implement and monitor "policies" and more specifically "public policies". The concept of "analytics" here is considered in the broadest possible way: I am personally interested to the concept of "reasons" (positive or negative) that enable a particular policy to be understood, explained, justified, argued, assessed; in other terms the construction of reasons thanks to which a public decision process becomes accountable. Within this subject I am interested in subjects such as:
- Modelling collective threats. The topic is at the intersection of two very broad areas. The first related to welfare economics, the issue of measuring the welfare of citizens and groups of citizens within the capability theory of A. Sen and the second related to adversarial risk analysis, where uncertainty of events could be guided by the strategic behavior of an agent considered to be an “adversary”. I extend findings to which I contributed in poverty measurement and urban quality measurement in order to build a general framework about the modelling of collective threats: events which may have a “negative impact” for communities and/or territories upon “commons” and public assets.
- Conflict management and transformation. The topic is related to my recent interest to Peace Studies. The central idea here is that conflicts are natural and potentially source of creativity and innovation in case the conflicting stakeholders manage to avoid violent confrontation and problem solving. My proposal consists in suggesting formal decision analysis tools in order to help stakeholders and analysts involved in conflict management. I am particularly interested in exploring systemic approaches allowing a deep “change” in the conflict.
- Policy Design. Most decision analysis tools are geared towards the assessment of public policies (ex-ante or ex-post), the set of such policies being already defined. It is crucial instead to become able to help decision makers in designing policies using formal models and tools. My interest consists in experimenting the ideas of alternatives design within the context of public policy as well as using formal design theory for this purpose.
Practice of Decision Aiding.
My research is not only theoretic. Research in Decision Aiding cannot be only theoretic. We regularly compare our findings in real life applications in order to refine our approach, to learn from successes and failures, get feedback from the interaction with the clients and test new ideas. Recent applications include (more details can be seen at the relevant publications in the papers page):
- Supporting Participatory Decision Making in Risk Management and Natural Resources Management.
- Urban Quality Measurement and Construction of Decision Maps.