Description du sujet de thèse

Objective: to understand, from models confronted with empirical data, how individuals' opinions and preferences interact with the structuring of social networks and groups to generate collective behaviors and social dynamics.

Keywords: computational social science, opinion dynamics, complex systems modeling, web-mining & deep learning, cognitive science.

Abstract: The formal modeling of social phenomena has known several interdisciplinary contributions since the founding work of John Nash on non-cooperative games with contributions from social psychology, sociology, cognitive economics or anthropology. Stylized theoretical models based on statistical physics and dynamical systems theory have addressed the central question of the articulation between individuals and the collective. These works, which formalized in a stylized way the concepts of opinions, norms and values, have only been confronted with real systems in a qualitative way, due to the lack of large-scale data on these phenomena.

The accessibility of massive data made possible by online social networks has changed this situation. Thus, and despite questions about the social segmentation of users of these platforms, it is nevertheless possible to use them to address several questions concerning social dynamics. For example, it is possible to study the formation of consensus, or polarization around a topic of social interest or to evaluate the phenomena of rumor or fake news diffusion.

This thesis project will aim at developing formalisms that can bridge different approaches to social modeling (game theory, statistical physics, dynamical systems, and complex networks) by integrating in a multilevel way the issues of opinion formation, values and norms, intrinsically linked to the personality of social actors, and the social structures in which they evolve (modeled as complex networks). In particular, we propose to study the dynamics of coupling between the processes of morphogenesis of the intrinsic properties of social actors and those of the interaction networks they form, through theoretical and empirical approaches.

The person recruited for this project will be in charge of developing new formalisms for the study of opinion dynamics, at the interface between statistical physics, game theory, agent models, complex network modeling and data science. These models will be fed by knowledge from the main theories of the individual (cognitive sciences) and the social (sociology/anthropology) and will be validated by relying on the solid mass of data and reconstructions from the macroscopes of the Institut des Systèmes Complexes de Paris Île-de-France (ISC-PIF) on topics such as French politics, climate and the COVID-19 pandemic.

This thesis is part of an interdisciplinary project funded by the 80PRIME call of the CNRS, which aims to better understand the long-term societal impacts of crises such as pandemics or global warming on our societies, by modeling the processes of collective attention, evolution of values and opinions and more globally, of change of interaction mode in society.

Contexte de travail

The Institut des Systèmes Complexes de Paris Île-de-France is a unit of the Centre National de la Recherche Scientifique, one of the largest and most prestigious French research organizations. The ISC-PIF is in partnership with more than a dozen other French universities and research organizations, as well as with the city of Paris and the Île-de-France region. The ISC-PIF is a place dedicated to the development of innovative and interdisciplinary research on complex systems at the crossroads of modeling, high-performance computing and megadata. Since 2005, it has catalyzed the emergence of common and interdisciplinary practices, facilitating access to skills, methodologies and the pooling of research resources. In addition to the CNRS staff who lead and animate research and development, the ISC-PIF has a scientific residency program that allows researchers from partner institutions to conduct projects while benefiting from the ISC-PIF's [platforms and services](https://iscpif.fr/services). The PhD student will benefit from an exceptional computing environment including grid computing, cloud computing (with Spark/Hadoop overlay), multicore servers (1TB Ram) and GPU clusters.

École Doctorale : formation doctorale « Sciences de la Société » co-accréditée EHESS / ENS-PSL.

Links : http://politoscope.org, http://climatoscope.org ; https://iscpif.fr/recherche/projets

Informations complémentaires

Education and skills expected: Master's degree in Applied Mathematics, Physics, Computer Science or related. Mastery of programming languages allowing the realization of extensive simulations and the analysis of massive data. Interest in an interdisciplinary approach.

How to apply: Send your CV, cover letter, transcript (L1-M2) and two letters of recommendation to the co-supervisors preferably before May 31, 2023. Selection will be based on an interview after selection on file by end of June 2023.