SMS Advertisement Assignment Problem with Meta Heuristic and Greedy Methods

Abstract

Advertisement is a very essential tool for marketing. SMS is a very popular medium used for advertisement since sending bulk SMS is very cheap. Advertisement companies usually have large subscriber pools and select interested people amongst them for sending SMS advertisements. SMS advertisement assignment problem (SAAP) is to match subscribers with advertising campaigns in order to increase ROAS (return on advertising spend). SAAP is special case of Generalized Assignment Problem (GAP) where weights for items are all same and equal to one. The target of SMS advertisement is usually huge number of people. So, approaches used for GAP is generally not quite suitable for SAAP. In this paper, two metaheuristic algorithms are proposed based on Genetic Algorithms (GA) and Ant Colony Optimization (ACO) which are used for GAP, as well. Experimental analysis demonstrates the limitation of these meta-heuristic algorithms due to curse of dimensionality (enormous search space for large problems). To overcome the size limitations two greedy algorithms are developed with a local search to improve the results. While meta-heuristic algorithms perform better than greedy algorithms for small problems, greedy algorithms produce better results in reasonable time for larger problem sizes.