By continuing to browse the site, you are agreeing to our use of cookies. (More details)

Doctoral fellowship (3 years) in computational social science

This offer is available in the following languages:
Français - Anglais

Ensure that your candidate profile is correct before applying. Your profile information will be added to the details for each application. In order to increase your visibility on our Careers Portal and allow employers to see your candidate profile, you can upload your CV to our CV library in one click!

Faites connaître cette offre !

General information

Reference : UMR8504-THOLOU-003
Workplace : AUBERVILLIERS
Date of publication : Thursday, May 07, 2020
Scientific Responsible name : Thomas Louail
Type of Contract : PhD Student contract / Thesis offer
Contract Period : 36 months
Start date of the thesis : 1 October 2020
Proportion of work : Full time
Remuneration : 2 135,00 € gross monthly

Description of the thesis topic

We are opening one doctoral fellowship (3 years) in computational social science, under the umbrella of the ANR-funded project RECORDS (2020-23) focused on the understanding of practices surrounding music streaming platforms. The fellowship focuses on the spatial dynamics underlying content consumption on streaming platforms, with music listening as a primary case study.

The largest music-on-demand streaming services are accessible in almost every country, through a standard interface, interoperable technologies, and almost identical subscription plans. A large part of platforms catalogs is common to all countries. As such they constitute a great benchmark to investigate the spatial dimension of online cultural consumption. While online information consumption has been heavily studied, its spatial dimensions are much less understood quantitatively. Individuals stream music on a daily basis, and like any other ordinary activity it is embedded in space through social contexts and habits. Applicants' research proposals may focus on different topics related to the project. Possible directions include (but are not restricted to):

- the role of context in individual consumption. The listening context (where you are, what you are doing, the people you are with) is assumed to determine the choice of content you listen to. However few results corroborate this hypothesis thanks to detailed, georeferenced individual activity data. Additionally,
recommender systems hardly integrate this information so far. Log files include low-level information (timestamp, ip, device, platform features) that are useful to infer higher-level spatial contexts thanks to statistical learning (e.g. 'at home', 'at work', 'on the move', 'partying'). Such higher-level information may be used to compare intra-individual and inter-individual variability, and suggest models useful to relate context, musical content and listeners satisfaction.

- the geographical diversity of tastes and practices on global platforms. If we spatially aggregate users streaming data and calculate the dominant music genres and artists in a given area, clusters will appear on the map. Along with the long established connections between social status/origins and taste preferences -- the geographical space being socially organized, we can expect to uncover spatial consumption patterns --, other factors may explain the observed differences between areas. They include local music scenes or the presence of cultural infrastructures. What are the geographical scales at which location explains music consumption? How strong remain local genres and traditions, including country-specific, on global platforms?

- the effects of external shocks/perturbations. How do users change their habits during extraordinary events? How useful are the activity data collected by platforms to retrieve higher-level information in such situations, e.g. human mobility?

- the spatial diffusion of songs, artists and genres. Many models have been proposed to capture the spatial diffusion of people, information, innovations, epidemics, etc. at different scales. How useful are these models to understand how songs and artists popularity grow through internet platforms? For example, a common strategy/constraint among artists to gain popularity is to move to a larger city ('if i can make it there i'll make it anywhere'). We could, for a limited number of cases, combine artists biographical data with georeferenced listening data spanning several years. How musicians trajectories relate to the number and location of their listeners on the internet? What are the respective contributions of social networks and space in shaping one's digital trajectory?

We expect a strong profile either in quantitative social science, or in computer science, applied mathematics or statistical physics. The work will more specifically involve modeling and data analysis skills (including spatial analysis/GIS at Géographie-Cités). A strong interest for interdisciplinary collaborations is mandatory.

Work Context

RECORDS is a collaborative research project funded through an ANR grant (2020-23). It involves several CNRS units, along with a partnership with Orange and with one of the major music on-demand platforms, Deezer. The goal is to improve our understanding of (i) the diversity of users practices and consumptions on streaming platforms (ii) the effects of manual and algorithmic content recommendation (iii) the space-time diffusion of music. This interdisciplinary research articulates quantitative and qualitative methods. It relies on an original data collection protocol that combines anonymized individual listening history data on the platform; detailed self-declared
information collected through a large-scale web survey; in-depth interviews with volunteer participants.
Project members come from different academic backgrounds, some from the social sciences (mainly sociology and geography), some from computer science and signal processing. The project is led by Thomas Louail (CNRS, Géographie Cités), Philippe Coulangeon (CNRS, Observatoire Sociologique du Changement), Camille Roth (CNRS, Centre Marc Bloch), Jean-Samuel Beuscart (Orange Labs SENSE) and Manuel Moussallam (Deezer R&D). More information (in french) is available on the project website, records.huma-num.fr

The PhD student will join Géographie-cités, a research unit located on Campus Condorcet, a social science research centre that opened in September 2019. It is located in Aubervilliers, in the inner northern suburbs of Paris, 20-minute away from the city center by public transportation. The campus hosts eleven research institutions and universities, more than sixty research units and thousands of people. The PhD student could also work part-time in Deezer R&D offices in Paris, 20 minutes away from the campus. She/he will be provided a laptop and secure access to all the sensitive data of the project that were collected either by the company or through the RECORDS websurvey. She/he will benefit from the project dynamics, which include possible collaborations with all the project participants (see the list on the project website). The thesis will be co-advised by Thomas Louail (PhD in computer science) and by another co-advisor, chosen according to the candidate's research interests.

We talk about it on Twitter!