Attached is a use case modeled somewhat after those
published on the DMTF site.
After spending time working this use case, I’m now
wondering if catalog “authoring” isn’t something we should consider for the
Cloud profile. For example, in the cloud use case (attached) we
have the following actors:
-
Customer – who brings his requirements to the
automated decision making process.
-
Stavros Investment Services – who brings its
know-how (recognizing patterns of symptoms & associated
actions)
-
Stock Quote Provider – who brings its
symptoms
In the current Spec, we don’t really provide any guidance
on how we take decision making requirements and map them to symptoms,
syndromes, and protocols. I could see an “authoring” flow working
something like the following:
-
Consumer “gets” Stavros’ SAF catalog to see what
Syndromes it recognizes.
-
The Stavros Syndromes could be “templates”, not
necessarily functioning Syndromes. For example, “Syndrome: Detect when
Quote {quote} price has increased {increase%} within
{timeinterval}.”
-
The Consumer could then instantiate such a “template”,
ie: Syndrome: Detect when Quote XYZ prices has increased 5% within 24
hours. It would be instantiated within Stavros’
Diagnostician.
-
The instantiation would also involve mapping Symptom
elements to the “template” Syndrome, in addition to the above hard
values.
Seems like Syndrome templates, Symptom mapping, and
instantiation are important aspects of SAF that we haven’t considered up to
this point.
Thoughts?