ParSeMiS - the Parallel and Sequential Mining Suite

Director:Philippsen, M.
Period:May 1, 2006 - December 31, 2011
Coworker:Wörlein, M.; Dreweke, A.; Werth, T.

The ParSeMiS project (Parallel and Sequential Graph Mining Suite) searches for frequent, interesting substructures in graph databases. This task is becoming increasingly popular because science and commerce need to detect, store, and process complex relations in huge graph structures.

For huge data that cannot be worked on manually, algorithms are needed that detect interesting correlations. Since in general the problem is NP-hard and requires huge amounts of computation time and memory, parallel or specialized algorithms and heuristics are required that can perform the search within time boundaries and memory limits.

Our target is to provide an efficient and flexible tool for searching in arbitrary graph data, to improve the adaption to new application areas, and to simplify and unify the design of new mining algorithms.

In 2010, the distributed stack implementations have also been tested on other algorithms and data structures.

watermark seal