Proceedings of the 15th International Conference on Information Integration and Web-based Applications & Services (iiWAS '13)
In recent years, dozens of publicly accessible Linked Data repositories containing vast amounts of knowledge presented in the Resource Description Framework (RDF) format have been set up worldwide. By utilizing the SPARQL query language, users can consume, integrate, and present data from a federation of sources for different application scenarios. However, several challenges arise for distributed query processing across multiple SPARQL endpoints, such as devising suitable query optimization or result caching strategies. For implementing these techniques, one crucial aspect is determining appropriate endpoint features. In this work, we introduce several metrics that enable universal and fine-grained characterization of arbitrary Linked Data repositories. We present comprehensive approaches for deriving these metrics and validate them through extensive evaluation on real-world SPARQL endpoints. Finally, we discuss possible implications of our findings for data consumers.