Added README.md

Author: Theomat

examples/pbe/transduction/knowledge_graph/README.md

@@ -0,0 +1,11 @@
+
+# How to reproduce the experiment?
+
+First you need a database that supports SPARQL queries.
+Once you have that, you can generate the database using the ``fill_knowledge_graph.sparql``.
+
+Then you need to execute ``convert_kg_json_tasks.py`` to convert ``constants.json`` to ``constants.pickle`` (supported by AutoSynth).
+Then you need to preprocess the tasks with ``preprocess_tasks.py`` which will guess the constants.
+Then you can use the ``constants.pickle`` file in ``evaluate.py`` with your model.
+
+In our paper the model was the one obtained through the scrip ``test_performance.sh`` in experiments.