A Fast Topological Parallel Algorithm for Traversing Large Datasets

This work presents a parallel implementation of graph-generating algorithm designed to be straightforwardly adapted traverse large datasets. new approach has been validated in correlated scenario known as the word ladder problem. The induces same topological structure proposed by its serial version and also builds shortest path between any pair words connected words. implemented parallelism paradigm is Multiple Instruction Stream - Data (MIMD) test suite embraces 23-word instances whose intermediate were extracted from dictionary 183,719 (dataset). morph quality (the two input words) performance (CPU time) evaluated against original algorithm. generated optimal solution for each tested, that is, minimum connecting an initial final was found. Thus, there no negative impact on solutions comparing them with those obtained through ANG However, outstanding improvement considering CPU time required build solutions. In fact, up 99.85%, speedups greater than 2.0X achieved