Data Provenance (DPS) TC

 View Only
  • 1.  Provenance Information Model

    Posted 23 days ago

    The meeting minutes mention "Discussion of using an information model as a central artifact from which different data encodings (JSON, XML, etc.) are derived will be taken up outside of this meeting". This post is wrt that discussion.

    I highly recommend members read https://docs.oasis-open.org/openc2/imjadn/v2.0/imjadn-v2.0.pdf (or it's html or md equivalents) about information modeling. Besides learning about information modeling, it's an example of a CN so you'll see the kind of boilerplate we need to add (and you can skim over the first 4 pages to get to the meat starting on page 5 and you probably only need to read to page 14 unless you really want to geek out).

    Aside with respect to the 'abbreviation' discussion on a sister thread - look at the 9 links at near the beginning of the doc and note this CN abbreviation was imjadn and it appears 18 times (hence my argument for brevity for our abbreviation, albeit we must be allowed more than 4 characters since imjadn is 6 characters).

    One aspect of information modeling that maybe isn't adequately covered in the CN is the advantage to 'spec writing' by the TC. Besides the consistency arguments made in the CN, there is also the fact that you can automagically create property tables to put in the spec. The current D&TA standard is sparse in that the text in the standard itself doesn't actually say the contents of the spec. It just 'standardizes' the JSON, XML, and YAML files. In reality, other than us true geeks, most people skip over reading the JSON, XML, YAML, (or the JADN if we were to include an information model). Whereas if we include the property tables in the spec itself, I believe we'd get 'more eyes' reviewing it. At least that has been my experience on other TC's in other specs. I would recommend the JADN be the definitive spec is the case of discrepancies, but having JADN-erived property tables make it much easier for humans to review.

    Of course agreeing to create an information model means someone has to actually do that. Since this TC has one of the world's leading experts on the topic (ie Dave Kemp), Im hopeful we'll get a contribution with info model content. Otherwise it's a moot point.

     



    ------------------------------
    Duncan Sparrell
    Chief Cyber Curmudgeion
    sFractal Consulting LLC
    Oakton VA
    703-828-8646
    ------------------------------


  • 2.  RE: Provenance Information Model

    Posted 20 days ago

    Since GitHub now has draft PRs, PR #10 https://github.com/oasis-tcs/dps/pulls contains a short summary of and rationale for using an information modeling process for DPS.  It's still sausage but should be ready to merge this week.  The original (easier to read than a PR) is https://github.com/davaya/dps/tree/information-model/contributions/InformationModel.

    Compared with the JADN committee specification https://docs.oasis-open.org/openc2/jadn/v2.0/csd01/jadn-v2.0-csd01.html and sections of the JADN committee note Duncan mentioned, this PR is short enough for an elevator speech, and it contains examples of JADN property tables usable for publication.