Dynamic Scheduling on Taxi Advertising Systems

In-taxi video advertising system is a brand new medium to promote various products and information by playing video advertisements in taxis. It introduces many interesting business models with potential profits in the digital advertising market. Each advertisement in the system specifies a play count requirement and its contract period. The system should finish playing the contractual advertisements by the end of their contract periods to meet play count requirements. However, due to the fact that the limited carrying time provided by all operating taxis is shared by multiple advertisements, unfair assignment of the play count may happen if we play advertisements in an inappropriate order during carrying services: Some of the advertisements may be played too much, so the play counts of others would be sacrificed. In this thesis, we apply the concept of utility function to represent the level of satisfaction of an advertiser for various advertisements, and propose an algorithm to schedule the display order of advertisements with fairness concern such that the total utility is maximized. We conduct a series of experiments to emulate the operations of the system in the real world with real data. The results show that our proposed scheme can achieve better performance on total utility and fairness for all advertisements in the system.