Greetings to everyone.
This is in regards to hearing suggestions on use-cases in "Graph or Node embeddings" for a fraud detection/prevention domain. I have gone through many blogs/videos on how embeddings can be helpful.
The fraud network we have basically involves connecting people using similar identities such as a phone or email to identify fraud. We use Neo4j.
There are some use cases that I have come across with respect to what we can do with graph embeddings.
- Visualization/pattern discovery
- Clustering groups
- Machine learning where models can use embeddings as one of the features etc.
I would like to export graph data out of neo4j into python and use inbuilt algorithms within the python libraries.
I want to get started with something that can be done usefully in the area of fraud detection but have no idea on how to get started with embeddings.
For a batch-processing architecture where we run jobs periodically to store people and their data and use inbuilt neo4j algorithms such as centrality and community detection, how can embeddings be usefully leveraged?
All suggestions could be greatly helpful and I can't wait to get started with embeddings!