Subgroup: Section 5.2, Type System Language Spec

[ehn@cs.cmu.edu]

Dear Subgroup Members: Thilo, Antonio, Adrian, Vinod and Junichi, 

It's time to get the discussion going on Section 5.2: Type System Language. Here are my comments on the section to kick things off. Other TC members are welcome to comment too!

-ehn

1. The section introduces several terms that could confuse a non-expert. Do 'type system', 'artifact metadata' and 'object model' refer to the same thing? If not, can we explain the distinction more clearly?

2. I think the way that Ecore is used is well-documented; the excerpt from Budinsky is a good idea. I also think the example in Figure 5 illustrates how Ecore gives you the representational power to model entities and relations that are useful for text analytics. What the spec doesn't discuss: Are there are any limitations associated with this choice?

3. The compliance point (p. 30) seems reasonable. But it doesn't mention the use of type system types to define input/output capabilities of an analysis operation. We might want to make this part of compliance as well.