Fig. 8From: Grasping frequent subgraph mining for bioinformatics applicationsSupport count of candidate graph g in a single graph G. a Non-overlapping versus overlapping graphs. For non-overlapping counts, only those instances of the subgraph are counted that are entirely unique and thus don’t share any of its nodes or edges with another instance. Overlapping counts allow subgraphs to share nodes and edges, as long as there is at least one node/edge difference with another subgraph (to distinguish between identical instances). b Non-monotonic example. Subgraph g1 has a support of 1, despite that the larger subgraph g2 that contains it has a support of 2. Subgraph g1 will only be counted once in graph G as no other instances is available that uses a different set of nodes or edges. This general procedure is equivalent for both multi- and single graph problemsBack to article page