W3C home > Mailing lists > Public > public-aikr@w3.org > May 2020

next steps Goals, Objectives

From: carl mattocks <carlmattocks@gmail.com>
Date: Sun, 24 May 2020 13:13:56 -0400
Message-ID: <CAHtonukof+eyNKdAyG81CCMvCh3ep_A91DjiSWPBit9Fi1XinA@mail.gmail.com>
To: W3C AIKR CG <public-aikr@w3.org>
Tuesday May 26th is the last scheduled meeting of the series focused on
Leveraging the StratML specification for Trustworthy AIKR  .. with a
Walk-the-talk focus of using StratML for AIKR,.

During our meetings we have benefited from the deep expertise of Owen ,
Chris & Jorge and agreed that :

   - AIKRCG Strategy can readily be defined using the Vocabulary and
   Template of the international standard StratML part 1
   - AIKRCG can use StratMl part1 & part 2 to EXPLAIN   how to 'Evaluate
   Trustworthy AIKR objects implemented by machine learning powered services'
   - AIKRCG will supplement the elements/attributes/tags StratMl XSD and
   XML with Knowledge objects that have KRId's to apply Knowledge-directed
   Artificial Intelligence Reasoning Over Schemas (KAIROS) DARPA-SN-19-19 .
   Towards that objective :
   -StratML is our Schema start point  for reasoning, as in, the
   performance of AIKR   inference
is scoped
   / weighed by the declared strategy.
   -AIKR reasoning uses KRID identifiers and data
   (aka metadata) properties, such as KR TYPE that has a value-set that
   includes ' Declarative'-, 'Imperative  (aka procedural)'.

based on these agreements we are confident that a near-term outcome of our
effort will include:

   - The AIKRCG Strategy which is published for human and machine
   - An AIKRCG demonstration, for humans and machines, explaining how an AI
   Strategist can produce a performance plan for AIKR objects implemented by
   machine learning powered services that are measured by Key Performance
   Indicators (KPIs)
   - An AIKRCG constructed core ontology (for human and machines) populated
   with the essential concepts and distinctions required for
   Artificial Intelligence Reasoning Over StratMl Schemas supplemented
by Knowledge
   objects with KRIDs

At the close of Tuesdays meeting we will discuss next steps -please reply
to this email if you have Goals, Objectives that should be discussed


Carl Mattocks

It was a pleasure to clarify

On Sat, May 23, 2020 at 10:30 PM Paola Di Maio <paoladimaio10@gmail.com>

> Carl
> all conversation and exchanges about these CG activities are carried out
> through the public mailing list
> (private exchanges are not part of the CG activities as such)  This is why
> meetings should be publicly announced
> on the mailing list and discussions/decisions documented in some form (I
> now understand that the meetings you are coordinating every other tuesday
> are about stratml adoption rather than about the CG activities in general)
> Look forward to learn more about what you have in mind for KRID
> pdm
> -
> On Sun, May 24, 2020 at 10:07 AM carl mattocks <carlmattocks@gmail.com>
> wrote:
>> Strongly suggest that the CG not be copied on one to one discussions.
>> Carl
>> On Sat, May 23, 2020, 9:56 PM Paola Di Maio <paoladimaio10@gmail.com>
>> wrote:
>>> Carl
>>> yes, we agreed to continue the discussion via email
>>> and I have posted the questions in an email to follow up our agreement
>>> but now I dont understand why Paul is replying - you mentioned KRID as
>>> your own contribution
>>> (if I remember correctly)  I would have expected the reply to come from
>>> you
>>> I have no preconditions on any topics, and I dont understand what is
>>> prompting your question
>>> what makes you ask?
>>> On Sun, May 24, 2020 at 9:52 AM carl mattocks <carlmattocks@gmail.com>
>>> wrote:
>>>> Paola
>>>> We have agreed to continue our discussions via email .. if you have
>>>> preconditions about what topics can be included please let everyone know.
>>>> Carl
>>>> On Sat, May 23, 2020, 9:43 PM Paola Di Maio <paoladimaio10@gmail.com>
>>>> wrote:
>>>>> Paul
>>>>>  simply trying to figure out what type of values you identify in the
>>>>> KRID according to what logic and schema
>>>>> and where (what domain) would that be applicable to and to solve what
>>>>> problem-
>>>>> I dont have an expectation as such -  but I figure anything that makes
>>>>> sense would do
>>>>> when I asked the question to Carl what exactly is KRID ( the KRID
>>>>> proposal emanated from Carl, so I expect Carl to send replies if this
>>>>> proposal comes from you, maybe you need to clarify that also)
>>>>> he said he would see a top level distinction between declarative and
>>>>> procedural
>>>>> i then sent an email pondering a few points about that and you reply
>>>>> Please start a document where you specify what is KRID and how you
>>>>> envision it to work
>>>>> then we can talk about it' at the moment, it is very difficult to have
>>>>> an intelligent exchange about it :-)
>>>>> pdm
>>>>> On Sun, May 24, 2020 at 9:32 AM carl mattocks <carlmattocks@gmail.com>
>>>>> wrote:
>>>>>> Paola et Al
>>>>>> Please outline your expectations for a taxonomy.
>>>>>> Carl
>>>>>> On Sat, May 23, 2020, 9:18 PM Paola Di Maio <paola.dimaio@gmail.com>
>>>>>> wrote:
>>>>>>> Paul-
>>>>>>> Thanks for reply
>>>>>>> *  In the parsing of a StratML XSD I found that:  *
>>>>>>> This is what the question is about -
>>>>>>> What did you parse the text with?
>>>>>>> Please share the parser and the output so that we can make better
>>>>>>> sense of your observations?
>>>>>>> pdm
>>>>>>> On Sun, May 24, 2020 at 4:19 AM Paul Alagna <pjalagna@gmail.com>
>>>>>>> wrote:
>>>>>>>> namespace hiccup2
>>>>>>>> <Paola> pls say how did you process the file- </
>>>>>>>> I'm not sure what you meant by your question
>>>>>>>> but it sounded to me like "how did you get to realize this?" So,
>>>>>>>> I'll answer that one.
>>>>>>>> Aside from the initial white space, the area of information for an
>>>>>>>> XML / XSD document is the beginning brace character "<" up to but not
>>>>>>>> including the next brace character "<"; inner split by a ">" token.
>>>>>>>> IE <stuff1>stuff2|"<" this "area of information" is also known as a
>>>>>>>> "fragment"
>>>>>>>> The XSD standard has rules about what information items are
>>>>>>>> contained in "stuff1" and "stuff2"
>>>>>>>> In "stuff1" attributes are recorded in the format
>>>>>>>> attributeName="attributeValue". If an attribute name is further split into:
>>>>>>>> namespaceName ":" localName then further processing is called for.
>>>>>>>> The XSD standard for namespaces says that a secondary XSD of that
>>>>>>>> namespace exists and that a workflow (XSD fragment) for the localName will
>>>>>>>> exist.
>>>>>>>> This is accomplished through 3 part mechanism:
>>>>>>>> 1-the namaspace XSD file is declared in the schema statement using
>>>>>>>> the "xmlns:" prefix such as <schema xmlns:foo="http//foo---" (oddly without
>>>>>>>> the .xsd ending )
>>>>>>>> 2- the namespace required is named in the attribute name AND/OR
>>>>>>>> value.
>>>>>>>> like: <element xsd:ref="foo:Fullname"
>>>>>>>> and
>>>>>>>> 3- that an XSD record exists in the namespace XSD:
>>>>>>>> IE <element name=localName-----..</element> existing in foo.xsd
>>>>>>>> In the parsing of a StratML XSD I found that:
>>>>>>>> The StratML.xsd calls for a stratml:Name and but the schema pointer
>>>>>>>> "xmlns:stratml=" does NOT point to a valid URI.
>>>>>>>> This is from the StratML.xsd itself
>>>>>>>> xmlns:stratml="urn:ISO:std:iso:17469:tech:xsd:stratml_core"
>>>>>>>> After a little digging I made the assumption that the usage
>>>>>>>> intended was to use the StratML.xsd as the secondary namespace XSD, in
>>>>>>>> addition to being the guiding XSD for stratML XML reports.  Because the
>>>>>>>> <element name="Name" XSD fragment does exist in this very document,
>>>>>>>> I can continue on. "I" can continue because I'm a human.
>>>>>>>> Any automatic processes like the AIKR information extraction tools
>>>>>>>> we are defining and building MUST follow the rules laid out by our
>>>>>>>> standards and the standards we dictate.
>>>>>>>> Paul
>>>>>>>> Thoughts? , comments?
>>>>>>>> Thanks
>>>>>>>> PAUL ALAGNA
>>>>>>>> PJAlagna@Gmail.com <PJAlagna@gmail.com>
>>>>>>>> 732-322-5641
Received on Sunday, 24 May 2020 17:14:47 UTC

This archive was generated by hypermail 2.4.0 : Sunday, 24 May 2020 17:14:48 UTC