Summary:

• Which datasets be can uses to test out lead extraction?
  • give the ability to parse selected articles across projects (custom provided datatset)
  • give the ability to parse entire category (ability set depth limit, for example max 100 articles)
  • give the ability to parse entire project if needed (ability set depth limit, for example max 100 articles)
• Testing against existing endpoints?
  • testing against summary looks like a good idea, in the beginning just figure out the way to put both of results back to back for human review
• Modelling: packaging: separate endpoint or fit it current APIs, repeat this exercise for all APIs?
  • I think it should be a separate API endpoint and a separate workflow, something like /v2/*-article where we replace * with something, main reason here is that this will expand our payload a lot and conceptually different enough
• Modelling: schema: is schema.org enough or do we need to expand?
  • we will need to expand on the schema in the future, but we are more than equipped to start with what schema.org currently provides