Finding Ad-hoc Similarity of a Node or set of nodes (without running similarity algorithm on all the nodes)

[mangesh_karangu]

I want to know if it is possible to find Similarity of a new Node (or a set of nodes) that were newly added to the graph.

To explain in mode detail.

• I have Jaccard Similarity calculated for nodes 1 to 100
• I then add nodes 101
• Is it possible to find similarity of node 101 to among nodes 1 to 100 without running similarity again on nodes 1 to 101

reason -

• say the similarity algo runs in batches every 24 hrs at midnight, but i need similarity calculated for new nodes between those 24hrs on an ad-hoc need
• a new node does not really exists but i just want to find the similarity

[Joel]

Hi Mangesh,

Documentation is here
https://neo4j.com/docs/graph-algorithms/current/labs-algorithms/jaccard/

Have you tried the targetIds or sourceIds parameters? The docs read like this might enable what you want to do, I haven't tried it myself, but perhaps all nodes in 'data' and the new nodes in 'sourceIds'?

Let me know if what you find out.