I've been dabbling in Neo4J, mostly related to possibilities related to financial fraud and related topics, for around 3 years now. It's been fun to see Neo4J move up the awareness ladder over that time. I've really never gone much beyond the fairly elementary level of queries and I'd like to stiffen up my cypher abilities. One of the things I've wrestled with is understanding how well Neo4J scales. So far, for 'largish' graphs (say 1 MM nodes), we've had better luck stuffing everything into something like Networkx.
I'm Julie from the Nashville area. I've been exploring using Neo4j related to insurance fraud.
Scaling issues have been a challenge for me as well. Our data tends to stream through, not the batch method most of the tutorials use. We're going to start exploring creating individual graphs for specific use cases to see if we can't get faster performance. That way we can remove extraneous relationships and target index just the things we'd use for that use case.
Hi Julie - thanks for the note. I'll keep my eyes out for your success stories.