Trajectory-driven Influential Billboard Placement
share
In this paper we propose and study the problem of trajectory-driven influential billboard placement: given a set of billboards U (each associated with a location and a cost), a database of trajectories T and a budget L, the goal is to find a set of billboards within the budget so that the placed ad can influence the largest number of trajectories. One core challenge is that multiple billboards have influence overlap on trajectories and it is critical to identify and reduce the influence overlap. Two more constraints on top of this challenge, i.e., the budget constraint and non-uniform costs associated with each billboard, make this optimization problem more intricate. We show that this problem is NP-hard and propose an enumeration based algorithm with (1-1/e) approximation ratio and O(|T|·|U|^5) time complexity, where |T| and |U| are the number of trajectories and billboards respectively. By exploiting the locality property of billboards influence, we propose a partition-based framework . partitions U into a set of small clusters, computes the locally influential billboards for each cluster, and merges the local billboards to generate the globally influential billboards of U. As a result, the computation cost is reduced to O(|T|·|C_m|^5), where |C_m| is the cardinality of the largest partition in U. Furthermore, we propose a method to further prune billboards with low marginal influence; significantly reduces the practical cost of while achieving the same approximation ratio as . Experiments on real datasets show that our method achieves high influence and efficiency.
READ FULL TEXT
