MULDER (Molecule-based Query Decomposition for RDF) is an adaptive federated query engine for federation of Web access interfaces. MULDER describes Web interfaces in terms of RDF molecule templates, i.e., abstract RDF graphs that characterize the accessible data. Moreover, MULDER utilizes molecule templates for selecting relevant Web interfaces, query decomposition, and execution. As a result, MULDER exploits the query features of SPARQL endpoints and Linked Data Fragments for obtaining complete answers for federated queries of arbitrary complexity.