In this talk I will describe a number of new techniques that allow the detection of an “emergent” relational schema from existing knowledge graphs (encoded with RDF) that are available on the Web. On a wide variety of datasets, we found that the detected structure explains well over 90% of the original graph. Furthermore, I will describe methods to give short names that humans find logical to these emergent tables, columns and relationships between them. These techniques can be exploited in many ways, e.g., to improve the efficiency of SPARQL querying systems, or to use existing SQL-based applications on top of any RDF dataset using a relational database engine.