Searching for Shortest Common Supersequences by Means of a Heuristic-Based Genetic Algorithm

In this paper we describe a genetic algorithm (GA) for the Shortest Common Su-persequence (SCS) problem which is a classical problem from stringology. The SCS problem has applications in artiicial intelligence (speciically planning), mechanical engineering and data compression. It is NP-complete even under severe restrictions concerning the alphabet size, the length of the given strings, or their structure. Using a Genetic Algorithm to solve SCS is not easy, e.g. because the search space contains only a few valid solutions of reasonable length and a natural representation would lead to varying string lengths. To circumvent these diiculties, we base the Genetic Algorithm on a slightly modiied Majority Merge heuristic. The resulting GA/heuristic hybrid yields signiicantly better results than Majority Merge alone and other well-known heuristics, though its running time is much higher.