In this work, methods are developed to overcome the inherent problems of network abstraction and analysis from multiple heterogeneous data sources. RDF and attributed graphs are two common choices for graph modeling. While both are very similar with respect to the type of information that can be represented, characteristics intrinsic to each representation affect the analysis performed over the resultant network abstractions. By selecting a dual graph representation approach and leveraging the strengths of both models, the semantic analysis performed over RDF graphs is combined with the topological analysis applied to attributed graphs, to produce a comprehensive foundation for network analysis that cannot be easily achieved, nor its value matched, by one representation independently.