The goal of Question Answering over Knowledge Graphs (KGQA) is to find\nanswers for natural language questions over a knowledge graph. Recent KGQA\napproaches adopt a neural machine translation (NMT) approach, where the natural\nlanguage question is translated into a structured query language. However, NMT\nsuffers from the out-of-vocabulary problem, where terms in a question may not\nhave been seen during training, impeding their translation. This issue is\nparticularly problematic for the millions of entities that large knowledge\ngraphs describe. We rather propose a KGQA approach that delegates the\nprocessing of entities to entity linking (EL) systems. NMT is then used to\ncreate a query template with placeholders that are filled by entities\nidentified in an EL phase. Slot filling is used to decide which entity fills\nwhich placeholder. Experiments for QA over Wikidata show that our approach\noutperforms pure NMT: while there remains a strong dependence on having seen\nsimilar query templates during training, errors relating to entities are\ngreatly reduced.\n