Add post-processed graph analytics from networkX

curtislisle commented 2 years ago

Does this PR close any open issues?

Closes # N/A

Give a longer description of what this PR addresses and why it's needed

This PR includes support for new API endpoints: nodes_stats, nodes_filtered, edges_filtered.
The first endpoint provides server-based calculation for betweenness centrality, node centrality, pagerank, and in/out node degree
The second and third endoint enable extraction of a subgraph from a large graph by specifying a minimal threshold for in or out degrees of nodes to be included. Current behavior returns some unconnected nodes. Will look into this...

Provide pictures/videos of the behavior before and after these changes (optional)

Are there any additional TODOs before this PR is ready to go?

Further testing by Jack.

TODOs:

[ ] Determine why nodes_filtered endpoint can return disconnected nodes

curtislisle commented 2 years ago

Second commit is to loo with Arango cursors to pull all nodes and edges. Previous behavior (first commit) stopped at 1000.

curtislisle commented 2 years ago

I replaced networkX with graph_tool. The interface is the same. Performance should be better on large graphs. I pushed changes to this PR without style fixes yet, so Jack could look at it. I will next address Roni's review comments before we decide to merge.

curtislisle commented 2 years ago

We opted to close this PR, in favor of running algorithms directly inside Arango via Pregel and AQL.

multinet-app / multinet-api