Paradata Model

[Daniel Rehak]

Steve and I have revised the proposed paradata model and updated the document:  http://goo.gl/Cq6im

(the full JSON examples at the end have not been fully updated).

The model was simplified to have 4 top level elements: actor, action, object, related objects

Actor, action, object can all be further refined with 4 subelements: description/attributes, measure. date, context

We opted for consistency and expressiveness in allowing these 4 subelements to be applied to the major elements, recognizing that you can express the same activity in different forms.

The model will be the topic of discussion on our call @ 16:00 Eastern on 9/22. We're looking for feedback and more examples to test the model. We hope to conclude the discussion this week so we can finalize the JSON and begin testing.

- Dan

--
Daniel R. Rehak, Ph.D.

ADL Technical Advisor
Skype: drrehak
Email:  daniel...@gmail.com
            daniel.r...@adlnet.gov
Twitter: @danielrehak
Web:   learningregistry.org
            adlnet.gov
            lsal.org
Google Voice: +1 412 301 3040
Tel:       +1 412 931 7317
Work:   +1 412 931 7319
Mob:    +1 412 805 7683

[joe hobson]

As discussed on the last design call, I've added examples of how we would use this model to publish paradata from the Brokers of Expertise system. For the most part, it was not difficult to fit our expressions into this model, and I think it will be easy for a consumer to make sense of the data. 

One area that was a little unsure of was when a user tags a resource. I added the paradata as a description subelement under the verb, but it looks like the preferred activity stream method is to add them as subelement properties of the object. I think either could be acceptable, but I'm curious which would be better for service processing, either on the node, or on a system consuming the data.

... .joe