Guaranteed delivery targeted display advertising occupies an important share in online advertisement market.Relevant research often assumes that the publisher knows the exact values or probability distributions of impression supplies,which is highly uncertain before decision making in practice,due to the changing social hotspots and other various factors.Assuming that only partially distributional information is known,this paper constructs robust allocation strategies under the worst-case scenario with an ambiguity set.A distributionally robust chance constrained model is constructed,aiming at optimizing penalty cost and fairness.After analysis,the model is transformed that can be directly solved by existing optimization solvers.A convex approximation iterative algorithm based on golden section search is further designed to improve the effectiveness.Finally,the numerical experiments are carried out.The results illustrate the effectiveness and stability of our strategy and the corresponding design algorithm.
NETL
NSTL
维普
万方数据
