Knowledge Graph (R)Evolution and the Web of Data

When querying knowledge over the Web we typically consider the Web of Data to be a static point of reference that is always available and that never changes. However, when actually running queries "in the wild", we encounter a broad range of problems; spanning from the (un)availability of entire knowledge graphs (and their SPARQL endpoints) to outdated references between knowledge graphs and beyond that, we are almost entirely missing out on the availability of previous versions of knowledge graphs and provenance metadata about them. This position paper discusses these issues in context and sketches some of the solu- tions to mitigate them. In particular, this paper first discusses approaches to keep knowledge graphs available for continuous and scalable querying and afterwards presents an approach that enables community-driven updates so that mistakes can be corrected or missing information can be added. Then, the paper highlights what we can learn from RDF archiving solutions to better support evolving knowledge graphs. And finally, the paper puts these aspects into perspective and provides an outlook to open challenges and future work.