Graph/Network Challenge

[StevenFrancis]

I have an interesting problem with a database - this is a psudo-graph database where differing tables are linked together via a single, central link table. What i need is to try and identify from this central table all the graphs which do not contain specific objects types - I am basically after a subset of graphs based on objects that they link.

Below is the basic structure of the link table:
From Obejct Type | From Object Referece | To Obejct Type | To Object Referece | Date of Link Link Reason

Now the main kicker on this is that the link table holds several hundred million links already - I intially only need a proof of concept on this so can pull down a subset that proves the process works but am not sure where to start with this in KNIME.

Any pointers?

[stelfrich]

Hi @StevenFrancis,

Without more details and ignoring the fact that there might be functionality in your database (if it’s to some extent graph-based), your table structure sounds like an “edge table”. I’d recommend taking a look at this example workflow that uses a similar input table (though from a file) to build up a network:

Create Network from Edge Table – tobias.koetter

From the created network, you should be able to use the Node Neighbor Extractor to get all the nodes (which are objects in your case) which are connected to a defined node.

Does that sound reasonable to you or am I talking nonsense?

Best,
Stefan