Graph databases are interesting to me because they store data in the same way our human minds do, by remembering topics (nodes) and making associations between them (edges).
I work in fraud detection for a large US bank.
I plan to master graph analytic techniques to solve problems in fraud and in unrelated humanistic fields such as environmental climate science, etc. I hope to share what I learn as I become more able.
Neo4j can detect fraud thru the relationships, directly or by inference, that exist between the accounts and transactions that fraudsters use and have. Furthermore various algorithms, such as community detection can find communities in seeming unconnected data.