Opportunities and challenges in partitioning the graph measure space of real-world networks

Abstract Based on a large dataset containing thousands of real-world networks ranging from genetic, protein interaction and metabolic to brain, language, ecology social networks, we search for defining structural measures the different complex network domains (CND). We calculate 208 all using comprehensive scrupulous workflow statistical machine learning methods, investigated limitations possibilities identifying key graph CNDs. Our approach managed identify well distinguishable groups confer their relevant features. These features turn out be CND specific not unique even at level individual The presented methodology may applied other similar scenarios involving highly unbalanced skewed datasets.