I use Neo4j because the data structure was a graph, based on the underlying research papert, so the project had to follow this concept. However, the research project doesn't take the full potential of Neo4j or an other graph database, because it doesn't apply any graph related algorithms. So the whole architecture could have been also modelled in a simple relational way (PostgreSQL/MySQL/etc.). However, Neo4j with its very easy to learn query language Cypher was in the end a very good choice, because other researchers were able to set up a new testing scenario also on the database level with much flexibility.