The GRAAL of carpooling: GReen And sociAL optimization from crowd-sourced data

•Carpooling problem as multiobjective optimization of number of cars and enjoyability.•Weights for the multiobjective optimization learned by means of a user study.•Extraction of enjoyability and mobility demand from data like Twitter.•Results on real world data, from the perspectives of the city, and the users.•GRAAL can save up to 57% of the cars, while the total enjoyability is up to double.•From the user perspective the total per-car distribution of enjoyability is increased.