A computer-implemented method of associating a node of a first graph with a node of a second graph, each of the first and second graphs comprise sets of nodes each corresponding to a physical entity having a physical geographic location and one or more node attributes associated therewith. The method includes identifying a subject node of the first graph, filtering out nodes of the second graph that are unrelated to the subject node of the first graph to identifying a first subset of candidate nodes, identifying one or more first level edge attributes associated with the subject node, the first level edge attributes characterizing a relationship between the subject node and first level nodes of the first graph adjacent to the subject node, and filtering out nodes of the first subset of candidate nodes having first level edge attributes that do not correspond to the one or more first level edge attributes associated with the subject node to identifying a second subset of candidate nodes.