Section: Application Domains

Computational Social Sciences

Several projects related to research in social science and humanities and/or research transfer have started in 2015 and continued in 2016:

  • Personal semantics (Gregory Grefenstette). In the current digital world, individuals generate increasing amount of personal data. Our work involves discovering semantic axes for organizing and exploiting this data for personal use.

  • Gregorius (Cécile Germain & Gregory Grefenstette). An application of semantic structuring and automatic enrichment of existing digital humanities archives.

  • Cartolabe (Ph. Caillou, Jean-Daniel Fekete - AVIZ, Gregory Grefenstette, Michèle Sebag). The Cartolabe project applies machine learning techniques to provide a visual, global and dynamic representation of scientific activities from large scale data (HAL at the moment).

  • AmiQap (Philippe Caillou, Isabelle Guyon, Michèle Sebag, Paola Tubaro). The multivariate analysis of government questionaire data relative to the quality of life at work, in relation with the socio-economical indicators of firms, aims at investigating the relationship between quality of life and economic performances (conditionally to the activity sector). This will be the topic of the Divyan Kalainathan's PhD, with emphasis on learning causal effect with novel causal discovery algorithms, in collaboration with post-doctoral student Olivier Goudet and researchers at Facebook AI research.

  • Collaborative Hiring (Philippe Caillou, Michèle Sebag). Thomas Schmitt's PhD, started in 2014, aims at matching job offers and resumes viewed as a collaborative filtering problem. An alternative approach based on Deep Networks has been started by François Gonard within his IRT PhD.

  • Within the U. Paris-Saclay Nutriperso IRS (Philippe Caillou, Flora Jay, Michèle Sebag), we start investigating the relationships between health, diets and socio-demographic features, with the ultimate goal of emitting individual recommendations toward a more healthy diet, such that these recommendations are acceptable.

  • Foodtech (Paola Tubaro, Philippe Caillou, Odalric Maillard). An application of agent-based modelling and machine learning to the study of labor conditions in digital platforms. Focus is on online services and mobile applications for food production, delivery, and consumption.

  • Sharing Networks (Paola Tubaro). Mapping the "collaborative economy" of internet platforms through social network data and analysis.

  • IODS (Wikidata for Science).

Significant challenges include some Big Data problems:

  • learning interpretable clusters from bottom-up treatment of heterogeneous textual and quantitative data

  • aligning bottom-up clusters with existing manually created top-down structures

  • building a unified system integrating the "dire d'experts".

  • merging heterogeneous data from different sources.

  • moving from predictive to causal discovery algorithms, in line with state-of-the-art research on causality.


  • Amiqap is funded by the ISN Lidex, with Mines-Telecom SES, RITM (Univ. Paris Sud) and La Fabrique de l'Industrie as partners.

  • The collaborative hiring study is funded by the ISN Lidex, in cooperation with J.P. Nadal from EHESS.

  • Cartolabe is funded by Inria, in collaboration between TAO and AVIZ.