Re: [learningreg-dev] LRMI format and Slice Resumption Token

[Jim Klo]

On Feb 19, 2013, at 9:55 AM, finke...@gmail.com wrote:

I have a few questions and I am hoping someone else has come across some of these before

1) Resumption Tokens are not working for me on slicing data.

ie https://sandbox.learningregistry.org/slice?any_tags=LRMI , then trying to use the argument resumption_token, even tried resumptionToken. Nothing seems to be working for me. Without keeping any_tags argument in there I get an error.

I'm not sure what might be wrong here however I did just discover a resumption token bug in OAI-PMH; but should be unrelated….

2) For the Khan examples for LRMI,
I see "payload_schema": ["schema.org", "LRMI", "application/microdata+json"]
But the payload is
"resource_data": {"items": [{"type": ["http://schema.org/CreativeWork"], etc }]}

And creative work does not use the same attribute names as LRMI. Might be missing something about what the payload is supposed to be for LRMI. My assumption was that payload items would be type "http://schema.org/LRMI" and the attributes would be that listed in the LRMI spec.

Note that data on sandbox is not production data, and may contain errors!!!

FWIW: I created those from documentation that I was collaborating on with SLC, AppliedMinds, and LRMI folks.  I've not seen the final draft, however the intent is that Schema.org would adopt the LRMI property extensions (into CreativeWork) which would technically make http://schema.org/LRMI obsolete..  I've been awaiting the final draft to commit this to gold.

Also some notes about resource data properties:

payload_schema is a list of "hashtags" that describe the resource, as it was recognized that not all data has formal schemas (and at the time JSON Schema was just a figment of someones imagination).

        payload_schema_locator is the URI for the actual schema.  Technically the Khan example you point out follows the schema for application/microdata+json http://www.w3.org/TR/microdata/#json which is a transformation of HTML Microdata (of which Schema.org defines specific elements).

So you're correct in that CreativeWork does not 'currently' have these properties but here's the actual latest draft which proposes these changes: http://www.w3.org/wiki/WebSchemas/LearningResources

3) This might be covered next week when update and delete are talked about. But trying to slice data for comparisons by identity=, we still have to verify the signature in document because anyone could post their documents with whatever signer they wanted. Is there anyway to slice by something ie (auth user, public key)  etc that we can know what comes back is ours? We would still verify the signature but it would be a small comfort to know that we are sorting through lots documents that we are pretty sure are ours..

There wasn't originally a plan for discussing slice, however part of this release will include a 'slice light' which reduces the number of combinations that slice is capable of accessing, while keeping most of the common uses of slice. Details on the differences are here: https://github.com/LearningRegistry/LearningRegistry/pull/236  Walt is documenting. :)

 

- Jim