Implement efficient random motif searching via neural subgraph matching

[rjurney]

Motif search for heterogeneous networks - especially temporal heterogeneous networks - has fundamental scalability challenges. Neural Subgraph Matching proposes a technique using graph representation learning and vector search called NeuroMatch. NeuroMatch is an efficient neural approach for subgraph matching.

The source code for NeuroMatch is at github.com/snap-stanford/neural-subgraph-learning-GNN.

FAISS and Distributed FAISS

If the code doesn't scale, is this something we could implement using FAISS and Distributed FAISS?

[rjurney]

@ThePigLA Some papers related to [Neural Subgraph Matching(https://arxiv.org/abs/2007.03092) are:

• Sub-GMN: The Subgraph Matching Network Model
• Interpretable Neural Subgraph Matching for Graph Retrieval
• G-Finder: Approximate Attributed Subgraph Matching

[rjurney]

@ThePigLA check it out - there is code!

https://github.com/snap-stanford/neural-subgraph-learning-GNN

[rjurney]

@ThePigLA I had the url for the code wrong. I edited it in place, but it is: https://github.com/snap-stanford/neural-subgraph-learning-GNN