Great article. As I understand it, you’ve stored the data in 2 types of tables; one of nodes and the other — the edges.

With the data being stored in relational database however, the query would have to be long in order to recursively go deeper into multiple hops from the original vertex. And as such, you’ve implemented a graph query API to workaround the problem.

My question is : is there are plan to migrate to a graph database or some other kinds of triplestore to simplify the above process? Or is the existing method sufficient or actually better, than a maintaining a triplestore/graph database.

I’m coming from RDBMS background, and I’ve used to store data in nodes and edges tables previously. Just wondering how was the experience having to maintain such architecture for system of such a large scale.

Thanks