Hi Wiseguy (great name!),

Thank you for reading and getting in touch - I appreciate it!

I have nothing against Neo4J, in fact I used it for a movie recommendation app a few years ago and like it, but it isn’t a triplestore so I didn’t include it.

Neo4j is a graph database but doesn’t store triples or quads (RDF). Additionally it has its own query language (Cypher) whereas triplestores have a standardised query language (SPARQL).

If you are interested, there was a recent benchmark between AnzoGraph and Neo4J here. You will see that AnzoGraph outperforms Neo4J significantly. This could arguably be down to the fact that they are running analytical queries and Neo4J is an OLTP database but the fact that triplestores are standardised allows me to use both an OLTP and an OLAP triplestore in conjunction with one and other. I can also switch them out very easily which is not the case if I was using Neo4J.

Saying that, Neo4J has benefits such as flexibility as it can store non-directed relationships between nodes for example.

I know that Cypher to SPARQL converters have been made etc… but if we are to have a world full of linked open data, this standardisation is important.

Hopefully this helps and thanks again for reaching out,

Angus

Machine Learning Engineer and Researcher at Wallscope. Studying a PhD in Artificial Intelligence. Contact details at http://addlesee.co.uk/

Machine Learning Engineer and Researcher at Wallscope. Studying a PhD in Artificial Intelligence. Contact details at http://addlesee.co.uk/