In this talk we introduce PieClam, a new way to analyze and understand complex graphs by grouping together nodes of similar connectivity in communities, where community membership is continuous and there are overlaps between different communities. Unlike previous methods, which only consider communities with strong inclusive connectivity, we introduce the concept of exclusive communities which indicate strong disconnection between nodes. By incorporating exclusive communities in our model, we prove that PieClam can approximate any graph with a constant number of communities that depends only on the required precision. This property is not satisfied if one only uses inclusive communities, as is done in traditional methods. Experiments show strong performance approximating synthetic graphs and the model is competitive with state of the art models in identifying anomalous nodes within graphs. Advisor: Dr. Ron Levie