Re: definitions, problem spaces, methods

Thank you
well summarized. some quick comments from me
- wikipedia is useful, but I have not looked at that page because the
content changes, and somethings things which are not true are considered
true and I dont have time to fix the universe ( ) however it can be useful
to provide a snapshot at a glance of KR. maybe we could/should help to
maintain that wikipedia pageon KR making sure it reflects what we are
learning here? whatever way people get to hear about Brachman and the long
history and importance of KR is good.
-  the framework for development of this CG is open for members to shape,
provided it fits the overall scope (which I am not sure is spelled clearly
enough) which is partly educational , ie, fill the gaps left by traditional
CS courses which only teach a few things about KR, in a very minimalistic
way, serve as an educational vehicle for those who want to approach the
topic and study it, understand what is happening the field of AI /ML and
figure out how KR fits in there and possibly, the application level - as
well as
advancing KR itself with new methods (I have been working on the latter
with a couple of publications and talks in recent years)
Anyone can fit into that framework depending on where they are in their KR
journey
- depending on who is going to work with what level of commitment in this
mission
there is scope both in simply creating a general resource of reference on
AI KR, AND in working towards new standards
we have not seen much work from anyone other than Owen on StratML and some
exploratory posts

On Sun, Nov 6, 2022 at 12:57 AM ProjectParadigm-ICT-Program <
metadataportals@yahoo.com> wrote:

> Thanks for putting things in a framework.
>
> Let's start at the beginning. The official term that best describes what
> we are trying to do is knowledge representation and reasoning for AI.
>
> See:
> https://en.wikipedia.org/wiki/Knowledge_representation_and_reasoning
>
> Here the name Brachman, mentioned by Paola pops up.
> He defines the core issues for knowledge representation:
>
>    - Primitives
>    - Meta-representation
>    - Incompleteness
>    - Definitions and universals vs. facts and defaults
>    - Non-monotonic reasoning
>    - Expressive adequacy
>    - Reasoning efficiency
>
>
> It is important to note that the issues of primitives, meta-representation
> and definitions and universals vs. facts and defaults are actually topics
> central to philosophy, psychology and more recently cognitive science and
> neuro science.
>
> They center on what is reality and how do we observe and perceive this.
> Here we deal with consciousness, mind and cognition and the constructs we
> use, being concepts and percepts in decision-making and reasoning.
>
> Because AI is being used in the physical sciences, medicine, bio and life
> sciences, humanities, engineering and in the practice of science itself, we
> must add to this list of Brachman the following issues:
>
>    - uncertainty principles for measurement in quantum mechanics, signal
>    processing and information theory
>    - bias in perception, decision-making and reasoning
>    - reality, the nature of reality itself and its relationship to the
>    mind (reasoning), and the
>    codified forms and rules of human thought and reasoning: logic,
>    available since antiquity; dialectics as a process of logical reasoning;
>    and semiotics which focuses on the epistemological properties of the
>    extant domain (see: Philosophy in Reality: Scientific Discovery and Logical
>    Recovery : https://www.mdpi.com/2409-9287/4/2/22)
>    - universal conceptual frameworks and interchangeable
>    conceptualizations using mathematical tools such as category theory,
>    (algebraic) topology etc.
>
>
> I have left out quantum physics and its application in quantum computing
> and the recent findings detailing quantum processes at play in the physical
> brain and how these might relate to processes underlying decision-making
> and reasoning.
>
> Also left out explicitly is the adequacy of natural language for
> conceptualization in knowledge representation and reasoning as it is
> implicitly covered in the core issues pointed out by Brachman.
>
> I think I have covered all the bases that need addressing for knowledge
> representation and reasoning.
>
> From this and previous posts in numerous discussions we should now be able
> to come up with an actionable list of general objectives and goals that we
> may define as broad or as narrow as we deem fit for what this AIKR CG wants
> to achieve, that we can format using StratML.
>
> Any industry or academic discipline specific application can be found by
> appropriate scope definition for selection of specific KR&R conceptual
> frameworks, methods and technologies.
>
> Milton Ponson
> GSM: +297 747 8280
> PO Box 1154, Oranjestad
> Aruba, Dutch Caribbean
> Project Paradigm: Bringing the ICT tools for sustainable development to
> all stakeholders worldwide through collaborative research on applied
> mathematics, advanced modeling, software and standards development
>
>
> On Friday, November 4, 2022 at 07:51:26 PM AST, Paola Di Maio <
> paola.dimaio@gmail.com> wrote:
>
>
> Dave asks for definitions, Adeel asks why is this CG called AI KR CG
> Milton now wants to see KR bible
>
> I ll comment quickly and take this opportunity to create a new diagram
> <https://docs.google.com/document/d/1lYn8-YUvIS_k1rezgqdR0DX2BLTRYpvkJpmvYjAEpng/edit?usp=sharing>
> since answering these questions can serve as reorientation also for folks
> who joined recently
>
> 1. Dave, definitions for KR are to be found in the many textbooks and
> papers, some of which are linked to our pages (need to reorganise pointers
> to resources). KR is an established field. I cannot answer all your
> personal questions that relate to the field. if you study the books,(take
> my course?) you'll find your answers. most graduate courses do not contain
> answers to new research questions, this is why we become researchers
> You need to study the references and pick a definition that fits what you
> are trying to do. The example you pointed to yesterday (some ML that
> identifies deepfakes) uses KR, thanks for bringing it up. Why dont you dig
> up the code and analyze and look into how this NN uses KR to solve the
> deepfake problem? That would be very interesting-
>
> 2.Adeel, this CG is called AI KR because it addresses the problems caused
> by lack of KR in ML. It was started because some years ago, AI went mad
> with NN/ML without KR and I started getting nighmares of a world run by
> unintelligent machines (you should have seen the papers in top conferences,
> I was in agony)  This CG pointed out that KR needs to be in the picture to
> make AI intelligence
>  We have succeeded. KR is now back in the AI/ML learning picture
> The topics we have discussed here are being picked up by researchers (yes,
> plagiarized and published under their own name) Universities have started
> looking for PhD students in the field of KR applied to AI/ML
> I am glad something has been achieved.
>  There are gaps to be filled, contributions to be made
>
> 3. Milton, the KR bible/s  have been written decades ago, then forgotten
> We may be able to add future chapters, KR now needs to be understood and
> practiced in the context of these powerful new technologies coming out from
> labs. NN coupled with KR (hybrid) are the way forward
> This is what people have been working on thirty years ago already
> Your idea to start relating the work of this CG to knowledge
> representation challenges is good, maybe we can find some use cases in
> other CGs (I am a member of a couple of other CGs and have shared on the
> list a few posts where the problems they were trying to address are clearly
> related to KR that we could look at
>
> PDM
>
>
>
>