A genetic algorithm for learning profiles in broadcasting applications on the Internet

In this paper, we investigate the adequation of evolutionary algorithms for learning a user's preferences during his interactions with a digital assistant. This study is done in the framework of "broadcasting" on the Internet. In our experiment, a genetic algorithm open-loop is used to globally optimise a selection of "channels" among a very large number of choices. We show that this approach could answer the problem of providing multiple optimal solutions without losing diversity.