Towards the Next Generation of Large-Scale Network Archives.

Both data and computer scientists need graph (network) datasets in the design, comparison, and tuning of important scientific results and practical artifacts. Despite the abundance of data in practice, freely available datasets are usually difficult to access, limited in size and diversity, and are collected in small static archives. This work presents our vision towards a next generation of graph data archives. Therefore, we formulate six key requirements to guide the design of such archives. We further propose GraphPedia, a prototype architecture that addresses these requirements, and provides a large collection of different graphs, in many different storage formats, rich metadata, advanced searching, and on-demand graph generation. Once the open implementation challenges are resolved, GraphPedia will become a dynamic meeting space for exchanging graphs.