Re: March 31 meeting : Leveraging the StratML specification for AIKR - KAIROS

The contents of Director Scherlis' presentation is now available in 
StratML Part 1 format at https://stratml.us/drybridge/index.htm#AIER2

As time permits, I may render the contents of the flow chart below as a 
value chain in StratML Part 2 format.  Since it has several recursive 
elements, it is an interesting use case.  It may highlight the need for 
a fourth stratml:Relationship type 
<https://stratml.us/references/oxygen/PerformancePlanOrReport20160216_xsd.htm#RelationshipTypeType>: 
Recursive.

I'm not sure I understand exactly what they mean by "Top-Level 
Schemas."  However, generally speaking, schemas mean increased maturity 
of the information being documented. 
https://en.wikipedia.org/wiki/Machine-readable_document With reference 
to the following objective, they may be essential for human 
understanding, much less trust and management:

    Objective 3.1: Understanding, Trust & Management
    <https://stratml.us/forms/temp-stratml/temp.xml#_64fb5eb2-7235-11ea-870b-c6940183ea00>
    - Enable human users to understand, trust, and effectively manage
    the emerging generation of AI partners

At least, like any other partner, AI "partners" should have performance 
plans and report their results in an open, standard, machine-readable 
format.

Owen


On 3/29/2020 4:34 PM, carl mattocks wrote:
> StratML folk:
> For reference at our March 31 meeting : Leveraging the StratML 
> specification for AIKR  .. with a Walk-the-talk  focus of using 
> StratML for AIKR , as in:
>
>   * AIKRCG will have  a Strategy to  'assess the value of StratML for
>     AIKR'
>   * AIKRCG will use StratMl to document how 'Evaluate Trustworthy AIKR
>     objects implemented by machine learning powered services'
>
> additionally I am presenting for discussion -
>
>   * AIKRCG will use a KAIROS approach for 'AIKR & StratML Schema
>     enabled Media Analysis of Performance Plans '
>
>
> Below is from March 24th online *FCW Workshop: AI  titled* *AI Powered 
> Systems: Expectations and Realities 
> <https://download.1105media.com/Custom/Workshops/2020/AI/Bill_Scherlis.pdf>*
> William L. Scherlis, Director, Information Innovation Office, Defense 
> Advanced Research Projects Agency   Tuesday, March 24th online *FCW 
> Workshop: AI. *
>
>
> image.png
>
> Carl Mattocks
> CoChair AIKRCG
>
> It was a pleasure to clarify
>
>
>
>
> On Sun, Mar 29, 2020 at 3:25 PM Owen Ambur <Owen.Ambur@verizon.net 
> <mailto:Owen.Ambur@verizon.net>> wrote:
>
>     I viewed the recording of this webinar and here are few points
>     made by Eric Forseter of Data Robot:
>
>       * A key issue is how to create the right roadmap
>       * The number of use cases is exponentially greater than the
>         number of data scientists.  So the process must be automated.
>       * The first step is to develop good use cases.
>
>     Having learned of the VA's National AI Institute, I converted
>     their about statement to StratML format at
>     https://stratml.us/drybridge/index.htm#NAII
>
>     These objectives are closely related to the mission
>     <https://stratml.us/carmel/iso/AIKRCGwStyle.xml#_63f01b52-83e9-11e8-9a9a-23c8e53a5ccc>
>     of the AIKR CG and the purposes
>     <https://stratml.us/#DefinitionPurposes> of the StratML standard:
>
>         Objective 1.2: Explainable AI
>         <https://stratml.us/carmel/iso/NAIIwStyle.xml#_66885f5e-71e1-11ea-9878-140f2583ea00>
>         - Emphasize explainable AI.
>
>         Objective 2.2: Partnerships
>         <https://stratml.us/carmel/iso/NAIIwStyle.xml#_6688621a-71e1-11ea-9878-140f2583ea00>
>         - Explore potential partnerships with federal and non-federal
>         entities.
>
>         Objective 3.2: Collaboration
>         <https://stratml.us/carmel/iso/NAIIwStyle.xml#_668863dc-71e1-11ea-9878-140f2583ea00>
>         - Forge collaborations.
>
>         Objective 4.1: Handbook
>         <https://stratml.us/carmel/iso/NAIIwStyle.xml#_66886512-71e1-11ea-9878-140f2583ea00>
>         - Create an AI Tech Sprint handbook.
>
>     Roadmaps, use cases, and handbooks are immature models for
>     performance plans.  To be effective, one way or another, they must
>     be fleshed out as actual performance plans in order to deliver the
>     desired results.
>
>     Pradeep, while NIH's RFI regarding the update to their strategic
>     plan makes no explicit reference to AI, it is implicit in this
>     cross-cutting theme:
>
>         Data & Technology
>         <https://stratml.us/carmel/iso/NIHSPF2020wStyle.xml#_49ef31de-6a5b-11ea-9f51-705d1a83ea00>
>         - Optimize Data Science and the Development of Technologies
>         and Tools
>
>     As well as this objective:
>
>         Objective 2.2: Resources & Infrastructure
>         <https://stratml.us/carmel/iso/NIHSPF2020wStyle.xml#_49ef2cb6-6a5b-11ea-9f51-705d1a83ea00>
>         - Support Research Resources and Infrastructure
>
>     The VA is obviously among the organizations with which NIH should
>     be efficiently and effectively collaborating and partnering, and
>     vice versa, including via the application of AI.  It would be good
>     if we could show them not only how the stratml:Relationship
>     elements but also StratML query services can be used to make such
>     opportunities salient.
>
>     It would be good if some small share of the unfathomable amounts
>     of money about to be thrown at the coronavirus could be devoted
>     toward worldwide application of the good practice set forth in
>     section 10
>     <https://www.linkedin.com/pulse/open-machine-readable-government-owen-ambur/>
>     of the GPRA Modernization Act (GPRAMA).  It is quite literally a
>     matter of life and death.
>
>     Owen
>
>     ........
>