Comparing approach & results

[trezbit]

I came across the Medium article & this repo and was very excited to see the potential in leveraging this approach in a project involving KGs specific to Cybersecurity domain (with focus on use-cases around compliance, NIST CSF framework & controls).
I am using GPT/OpenAI (via prompts derived from published NIS/CIS relational maps) and have just began implementing the KG extraction/build module.
I am looking forward to test this generic KG builder against my current (more domain specific & discriminative) approach and see if both efforts can benefit from the findings.