Applying parallel DBMS for very large graph mining

Graph partitioning is an interesting topic in graph mining, that comes into use for some theoretical and practical problems (graph coloring, integrated curcuit desing, finite element modeling, etc.). The existing serial and parallel algorithms suppose that the graph being analyzed can fit into main memory along with all the intermediate data, so they cannot be applied for very large graphs. We introduce a new way of partitining - using the parallel relational DBMS PargreSQL that is based on open-source PostgreSQL DBMS.