I am thinking of applying Graph Neural Network (GNN) for anomaly detection. It seems that at the moment the best technique so far is BWGNN that's originally proposed here [1].

Another question is that you claimed querying SQL as graph without ETL. Do you use similar technique to associative array mathematics proposed by D4M or other techniques [2].

[1] "Rethinking Graph Neural Networks for Anomaly Detection" in ICML 2022:

https://github.com/squareRoot3/Rethinking-Anomaly-Detection

[2] D4M: Dynamic Distributed Dimensional Data Model:

https://d4m.mit.edu/

Thank you so much! Really insightful. I think we can definitely integrate these into PuppyGraph.