Location based advertisement (LBA) is a new contextual marketing tool which enables advertisers to attract customers more effectively than ever, by sending location-aware advertisements, which we call e-coupons, through mobile phones. LBA exploits the real-time location information of customers as well as demographic and personal traits. In LBA, due to the privacy and technology issues, an intermediate advertiser, which we call the super mobile, sends e-coupons to customers on behalf of stores.

more »

... We study the e-coupons strategy problem in LBA from the point of view of the super mobile. First, we show that the full information model of the problem is reduced to a well-known optimization problem, the multidimensional 0-1 Knapsack problem (MKP). MKP belongs to the NP-Hard class of problems. Because of the total unimodularity of the problem, we show that, in a special case, significant e-coupons strategy problems can be solved in reasonable time. Second, we develop dynamic models of the e-coupons strategy problem such as the online, semi-online, and dynamic and stochastic models. In the online model, we show that our problem is reduced to the online MKP and propose threshold type algorithms which use the weighted efficiency as a criterion to select items. E-coupons strategy problems do not have to be solved purely online, where a decision is made immediately upon the arrival of an item, the decision is not revocable, and the item cannot be reconsidered; therefore, we define the semi-online model which lies between the online and full information model. We show that a semi-online solution is better than an online solution in our experiments. In the online and semi-online models, the number of customers who visit the target area is assumed to be known a priori. However, in the dynamic and stochastic model, we assume that the number of customers visit the target area is a random variable drawn from a Poisson distribution and the amount of time they stay in the area is a random variable drawn from an Exponential distribution.