Building an open data portal for one of the main police forces in France.
Computational appraisal of gender representativeness in popular movies. (Review by Le Monde)
Workshop gathering researchers, public servants and entrepreneurs involved in the quantification of various case of discriminations using methods spanning from testing to machine learning.
Website (in French)
An analysis of topological confinement of recommendations on YouTube.
80 years of AI research represented as connectionist (neural nets) vs. symbolic (rule-based).
Large-scale diversity estimation through surname origin inference. Application to various datasets in France. (Review by Le Monde)
A geographical mapping of Google's suggestions made for different world regions, showing "a world of Google searches" (New Scientist).
An attempt to study "styles of reasoning" specific to machine learning algorithms through their history and usages. Scientific usages in various disciplines are observed with some analysis of WebOfScience data. Engineering usages are observed with data from Kaggle and StackExchange.
Manuscript (in french)
"When Big Data meets Porn" (The Atlantic) or "A Googe Trends for niche sexual interests" (Wired). More formally, a statistical analysis of millions of porn videos meta-datas, showing, among other things, categories trends and communities.
Co-founder and head of Fabelier, a hackerspace in Paris. More than 100 workshops and a community of 400 people gathered around Web, Electronics, Neurosciences, Hacktivisme and much more.
History and analysis of F/LOSS as a social movement and the cultural significations of hacking. Various analogies with programming are explored: reglementation, architecture, art. Finally, a typology of the inner logics of trangression and collaboration of this social movement is described.