Federated query processing. Federated query processing allows to execute queries at the web scale. For example, “What is the number of men and women famous scientists per birth year ?” can be expressed as a federated query. Source selection plays an important role during federated query processing, it allows to locate relevant knowledge graphs for each triple pattern of a query. During query execution, subqueries are generated and evaluated at the SPARQL endpoints of relevant knowledge graphs. Intermediate results are aggregated locally and the final query result is returned to the user. In DeKaloG, we aim to use the index to perform efficient source selection at the web scale. Transparency of the index should allow to remove duplicate sources. The index may also be used to assist users in writing federated SPARQL queries. For example, given a prefix or a namespace, a tool may provide a list of candidate properties or classes, implement completion of property or class names (as in ), and provide hints on possible or impossible joins of triple patterns.