Multi-agent systems in marketing: modeling with Petri nets.

Summary This thesis aims at modeling interactive multi-agent systems on the Internet. In a first part, we study the concept of interactive multi-agent systems through a large literature review. We then present Petri nets that we apply to two application cases. First, we model gift lists on the Internet. We evaluate the performance of these complex dynamic systems and then optimize the impact of e-mailing campaigns by simulating several marketing strategies. Our second application is a viral marketing game. We detail each block of the model used to describe all the steps of the process and then optimize the game through simulations (recruitment, number and order of questions). We conclude the thesis by detailing the advantages of Petri nets, the limitations of the work done and the set of future research avenues opened by the potential application of Petri nets to the marketing domain.