a bit more on World model versus ontology discussion from last weeks Ontology Summit discussion

[Gary Berg-Cross]

At the last ontology summit meeting  as part of my briefing I had a small slide on world models.  This leveraged that Gary Marcus had made about their role in making modern AI systems successful. One question was whether there was a difference between an ontology and a world model and I offered a simple observation that world models were typically more inclusive and dynamic. Some follow up remarks were made by Mike Gruninger and Deborah McGuinness about the difference between T box and A box parts of ontologies that needed to be taken into account. 

Mike had more to say about handling processes with formal representation.

There has been some follow up on the forum about the difference between knowledge bases and ontologies, but I think it's worth some comments on world models.

Some recent proposed advancements in Generative AI (GenAI) architectures emphasize the integration of a "world model" within their frameworks. Broadly this concept reflects an interest in deeper contextual understanding and situational awareness, in the hope of enabling AI systems to perform tasks more effectively by incorporating knowledge of the world they operate in. There is also now often a use in simulations, but in GenAI reinforcement learning, and generative tasks.

An example of the knowledge/awareness in such models might include geographical knowledge (say via a knowledge graph), but also social dynamics, and contextual relationships between entities.

Perhaps, just for operational reasons the way we typically think of ontologies, is as more static and therefore less a part of the operational system runninga world model. But of course the structure of the world model could be highly dependent on ontologies for the formal specification of concepts and relationships within a large world "domain".

So one difference that might be argued between them is one of purpose:

A world model is  used to simulate real-world dynamics and aid in decision-making for tasks while an ontology's purpose is to
provide a structured framework for knowledge sharing and reasoning. Typically, and historically, an ontology is more static, detailing specific relationships and definitions useful for interoperability while world models have tended to be more practical about the algorithmic dynamics, integrating numerous variables and relationships.

Gary Berg-Cross 

Potomac, MD

240-426-0770

[Alex Shkotin]

Gary,

Unfortunately, you don't provide links to either Marcus's talk or your own. This means it's impossible to quickly delve deeper, with just one click. Let me answer from a general perspective.

The question for genAI is: does it have a model of the world?

The question for IT-ontology is: does it contain a model of the world?

In both cases, the number one question is: what world are we talking about?

If it's a small world, like the ISS or Tiangong, then knowledge about it can fit in both genAI and IT-ontology.

Let W be some world. For example, it could be the world of a coffee shop for a robot waiter.

What are the criteria for IT-ontology, genAI, or robot-software to have a model of the world W?

That's the question.

Antithesis: IT-ontology can contain a model of the world.

Alex

вт, 10 мар. 2026 г. в 18:43, Gary Berg-Cross <gberg...@gmail.com>:

--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CAMhe4f0ifp7E%3DZErpT4sRtx099imLGjjVPRK2%2B7u6wK%3DBtbsHg%40mail.gmail.com.

[Marco Neumann]

Here is a link to the slide Gary Berg-Cross mentioned.

https://www.youtube.com/watch?v=iD9LhJ1PVwA&t=3375s

I am trying to capture the context here as a general GenAI criticism
as formulated by Gary Marcus, who argues that current LLMs lack true
"world models" - as in they do not understand physical, social, or
causal reality.

But a "world model" is not the same as talking about an ontology
exclusively, while Gary Marcus may refer primarily to logical
representations in the form of an ontology or multiple ontologies. or
more general in model theory (MT) terms, with a semantic theory of
truth and intention of providing a rigorous, logical representation of
how language relates to the world.

Janet Singer is hinting at that difference in the discussion here
https://youtu.be/iD9LhJ1PVwA?t=4613

And I think for Deborah McGuinness, GenAI "world models" are primarily
a competing theme that she fears might consume ontology solutions
space as part of knowledge graph offerings, as they are not
sufficiently probability or uncertainty-aware yet, and a strength of
GenAI solutions such as LLMs.

https://youtu.be/iD9LhJ1PVwA?t=4819

Best.
Marco

> --
> All contributions to this forum are covered by an open-source license.
> For information about the wiki, the license, and how to subscribe or
> unsubscribe to the forum, see http://ontologforum.org/info
> ---
> You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CAMhe4f0ifp7E%3DZErpT4sRtx099imLGjjVPRK2%2B7u6wK%3DBtbsHg%40mail.gmail.com.

--


---
Marco Neumann

[Alex Shkotin]

Marco,

Thank you. The beauty of IT-ontology is its formal representation of any knowledge. Moreover, "formal" simply means "mathematical." Mathematical representations of knowledge about the world are highly developed in physics and other sciences and technologies. IT-ontology formalizes the texts that accompany scientific and engineering knowledge using mathematical logic.

At the same time, IT-ontology can easily contain any mathematical formulas and algorithms. And, of course, any facts (A-box) about a given world.

On the other hand, genAI is a trainable black box. And if we say "there's no model of the world," the trainers respond, "We'll teach it right now."

And as I wrote, "What are the criteria for IT-ontology, genAI, or robot software to have a model of the world W?"

We can take these criteria from the GB-C slide[1] or from GM text.

And see if they can be implemented in IT-ontology.

I believe so.

From the model theory perspective you mentioned, I believe finite models will suffice, but somewhere nearby will be algorithms. May be in Wolfram system 📯

Alex

[1]

ср, 11 мар. 2026 г. в 14:40, Marco Neumann <marco....@gmail.com>:

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CABWJn4RwLHYQ84uPeraygQu-zi0iZvq1qk6wM6NFXdvWTC61ZA%40mail.gmail.com.

[alex.shkotin]

IN ADDITION: this article may be interesting https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/

четверг, 12 марта 2026 г. в 12:47:37 UTC+3, Alex Shkotin:

[Marco Neumann]

Thank you, Alex. I am currently approaching this from a review perspective on RDF Semantics [1]
And here in particular, its foundation on FOL and Tarskian model-theoretic semantics as edited by Pat Hayes with interpretations to a single possible world.

When the plural is used in "world models" in GenAI, what is possibly referred to here? Is it just a collection of potential incongruent truth sentences or simply a referent to a computational artifact, a massive collection of learned parameters (weights and biases) that approximates a statistical distribution? 

Best,

Marco

[1] https://www.w3.org/TR/rdf-mt/

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CAFxxROQc86RoPg1YnY2xQheTyXuGiogsdiKsDDLQoKZeAODwHw%40mail.gmail.com.

--

---
Marco Neumann

[Alex Shkotin]

Marco,

It is interesting that in your version of the document (2004) they have "world" 31 times. But in newer version (2014) - 0 times.

And it is interesting that in the former they have in Glossary:

"World (n.) (with the:) (i) The actual world. (with a:) (ii) A way that the actual world might be arranged. (iii) An interpretation (iv) A possible world.

(The metaphysical status of 'possible worlds' is highly controversial. Fortunately, one does not need to commit oneself to a belief in parallel universes in order to use the concept in its second and third senses, which are sufficient for semantic purposes.)"

Anyway the main question is not a math object (they use Model theory), but reference to reality from this math object.

As for genAI, my knowledge is very superficial.

Best,

Alex

чт, 12 мар. 2026 г. в 15:31, Marco Neumann <marco....@gmail.com>:

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CABWJn4QdhuSX4SNj%3DLS0s%2B7Uv8BBWT-hpNigA03v5ZvZZ0%2BLLw%40mail.gmail.com.

[John F Sowa]

Alex,

Formality implies logical precision.  That requires a very narrow notation that is much more likely to be FALSE than a readable summary in an ordinary natural language.

That fact explains why  WordNet had been the most widely used starting point for ontology for many years.  It also implies that WordNet is still an excellent, broad-coverage resource.  For any subject represented in any natural language, WordNet is far more likely to be a useful and workable starting resource than a formal ontology represented in a formal notation.

I am not saying that WordNet is the best or only starting point.  There are many other resources available for mapping ordinary language to and from various formal notations and systems.  But an application of those resources and supporting methods is essential before a formal representation can be developed. 

These issues, by the way, are topics that Doug Lenat and I had been discussing and working on for many years.  Although we used different resources and methods, we agreed that a precise formal statement comes at the END of the development, never at the beginning.

I also agree with the comments that Janet Singer had discussed.  We had been discussing related issues during many sessions when Ontolog sessions were widely attended in person.

There is much more to say about these issues, and that's a topic that Arun and I will be discussing in a later session of the Ontology Summit.

John

 

---

From: "alex.shkotin" <alex.s...@gmail.com>

[Sheth, Amit]

thought I would share my use of Worldmodel[TM[] during 1999-2001 when we build the first commercial knowledge graph/world model powered by semantic search engine

with this very learned group of colleagues.

Details on how the Worldmodel was built 9with continuously extractors to bring in updated information

about the digital world and keep the  the world model updated is in the patent and 2002 papers 

summarized/referenced here; https://www.linkedin.com/pulse/15-years-semantic-search-ontology-enabled-amit-sheth

Both types of representation of the world is needed.

Amit Sheth LinkedIn | Google Scholar | Web | Wikipedia | X | Quora NCR Chair Professor, CSE; Founding Director, Artificial Intelligence Institute University of South Carolina AIISC: LinkedIn| Web | Projects | YouTube

[Alex Shkotin]

Amit,

Judging by the "the "WorldModel" is defined as an internal representation consisting of ontologies, domain-specific knowledge, and rules used to process digital media and enhance search." here you think the same as Gary, who started this thread.

I am still thinking that we can keep "domain-specific knowledge, and rules" in IT-ontology. Taking into account for example the definition of Nicola Guarino as most comprehensive.

As for the idea itself, we should at least start with Laplace: "Pound, J. (trans.) (1809) The System of the World, 2 vols, London: Richard Phillips

• _ The System of the World (v.1)

• _ The System of the World (v.2)

"

Alex

пт, 13 мар. 2026 г. в 07:18, Sheth, Amit <AM...@sc.edu>:

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/C805EF39-F4D5-44F4-94B2-B6BB14CAB4B2%40sc.edu.

[Alex Shkotin]

John,

Any language, even natural language, is a rough approximation of reality, designed to capture the essence, i.e., to abstract correctly.

On the other hand, English is a fairly formal language. See (PDF) English is a HOL language message #1X.

Of course, formal representation is the final phase of knowledge processing. First, knowledge must be systematized and concentrated. Systematization consists of extracting the most important units of knowledge from the text: definitions, axioms, hypotheses, proofs. Concentration consists of storing these units in frameworks in the languages in which people were able to formulate them.

Here's an example of an axiom in a framework.

https://docs.google.com/spreadsheets/d/1gThpt1P8iulIl5SXfoekUiMg-562oiLfhHQYQjjg_y4/edit?gid=0#gid=0

rus	Пусть A, B - точки. Если A и B различны то существует единственная прямая которая соединяет A с B.
eng	Let A and B be points. If A and B are distinct, then there exists a unique straight line that connects A to B.
yfl	∀p1,p2:Po p1≠p2 → (∃1l:SL connects(l p1 p2)).
coq	line_existence : ∀ A B, A ≠ B → ∃ l, Incid A l ∧ Incid B l; line_uniqueness : ∀ A B l m, A ≠ B → Incid A l → Incid B l → Incid A m → Incid B m → EqL l m;

Note that all languages are equal in a framework. Modern ontologies written in OWL2 contain all other languages in annotations, which is logically incorrect.

Anyone who needs definitions specifically in OWL2 should simply extract them, in the required quantity, from the framework.

On the other hand, any text (textbook, article, presentation) using terms from the framework should reference the framework, or better yet, directly the knowledge units.

Modern, separately stored ontologies are knowledge references.

Alex

чт, 12 мар. 2026 г. в 23:48, John F Sowa <so...@bestweb.net>:

--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/bab17d820d2a4a69aabe06ad5f6b4b61%409325347c7411478abc092e9cf752bca3.

[do...@foxvog.org]

WordNet is a useful starting point for creating an ontology from scratch
since it includes terminology for a huge number of concepts and has
broader term - narrower term relationships included.

But if one searches a little, one may find another source that has terms
with better definitions or better inter-term relationships already
defined. In the medical field there are ICD-10, ICD-9, and other such
term sets. Other fields have similar curated term sets.

Basic concepts in most fields can also be found in freely available
OpenCyc, released some time ago
(https://standards.clarin.eu/sis/views/view-spec.xq?id=SpecOpenCyc )
with almost 240,000 terms. The OpenCyc knowledge base has subtype and
instance of relations for all its terms. as well as other useful
assertions that can be selected from and added to.

-- doug foxvog

> -------------------------

>
> From: "alex.shkotin" <alex.s...@gmail.com>
>
> IN ADDITION: this article may be interesting
> https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/
>
> четверг, 12 марта 2026 г. в 12:47:37 UTC+3, Alex
> Shkotin:
>
>> Marco,

>> Thank you. The beauty of IT-ontology [1] is its formal

>> representation of any knowledge. Moreover, "formal" simply means
>> "mathematical." Mathematical representations of knowledge about the
>> world are highly developed in physics and other sciences and
>> technologies. IT-ontology formalizes the texts that accompany
>> scientific and engineering knowledge using mathematical logic.
>>
>> At the same time, IT-ontology can easily contain any mathematical
>> formulas and algorithms. And, of course, any facts (A-box) about a
>> given world.
>>
>> On the other hand, genAI is a trainable black box. And if we say
>> "there's no model of the world," the trainers respond, "We'll teach
>> it right now."
>>
>> And as I wrote, "What are the criteria for IT-ontology, genAI, or
>> robot software to have a model of the world W?"
>>
>> We can take these criteria from the GB-C slide[1] or from GM text.
>>
>> And see if they can be implemented in IT-ontology.
>>
>> I believe so.
>> From the model theory perspective you mentioned, I believe finite
>> models will suffice, but somewhere nearby will be algorithms. May be
>> in Wolfram system 📯

>> Alex [2]
>>
>> [1]

>
> --
> All contributions to this forum are covered by an open-source license.
> For information about the wiki, the license, and how to subscribe or
> unsubscribe to the forum, see http://ontologforum.org/info
> ---
> You received this message because you are subscribed to the Google
> Groups "ontolog-forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ontolog-foru...@googlegroups.com.
> To view this discussion visit

> https://groups.google.com/d/msgid/ontolog-forum/bab17d820d2a4a69aabe06ad5f6b4b61%409325347c7411478abc092e9cf752bca3
> [3].
>
>
> Links:
> ------
> [1]
> https://docs.google.com/document/d/1N6vKV2FUly17U-gS7Sgzc67BxZUZz7h7mr6GJC9KexU/edit?tab=t.0
> [2] https://www.linkedin.com/in/ashkotin/
> [3]
> https://groups.google.com/d/msgid/ontolog-forum/bab17d820d2a4a69aabe06ad5f6b4b61%409325347c7411478abc092e9cf752bca3?utm_medium=email&utm_source=footer

[do...@foxvog.org]

The difference as i see it is that an ontology is a term and rule set
with the terms defined by type and rules expressing what can logically
be determined from sets of assertions matching certain patterns. A
world model provides data about individuals using ontologies. Different
world models can be valid at different times or in different contexts.
Different world models can be mutually inconsistent.

-- doug foxvog

On 2026-03-12 08:31, Marco Neumann wrote:
> Thank you, Alex. I am currently approaching this from a review
> perspective on RDF Semantics [1]
> And here in particular, its foundation on FOL and Tarskian
> model-theoretic semantics as edited by Pat Hayes with interpretations
> to a single possible world.
>
> When the plural is used in "world models" in GenAI, what is possibly
> referred to here? Is it just a collection of potential incongruent
> truth sentences or simply a referent to a computational artifact, a
> massive collection of learned parameters (weights and biases) that
> approximates a statistical distribution?
>
> Best,
> Marco
>
> [1] https://www.w3.org/TR/rdf-mt/
>
> On Thu, Mar 12, 2026 at 9:47 AM Alex Shkotin
> <alex.s...@gmail.com> wrote:
>
>> Marco,

>> Thank you. The beauty of IT-ontology [1] is its formal

>> representation of any knowledge. Moreover, "formal" simply means
>> "mathematical." Mathematical representations of knowledge about the
>> world are highly developed in physics and other sciences and
>> technologies. IT-ontology formalizes the texts that accompany
>> scientific and engineering knowledge using mathematical logic.
>>
>> At the same time, IT-ontology can easily contain any mathematical
>> formulas and algorithms. And, of course, any facts (A-box) about a
>> given world.
>>
>> On the other hand, genAI is a trainable black box. And if we say
>> "there's no model of the world," the trainers respond, "We'll teach
>> it right now."
>>
>> And as I wrote, "What are the criteria for IT-ontology, genAI, or
>> robot software to have a model of the world W?"
>>
>> We can take these criteria from the GB-C slide[1] or from GM text.
>>
>> And see if they can be implemented in IT-ontology.
>>
>> I believe so.
>> From the model theory perspective you mentioned, I believe finite
>> models will suffice, but somewhere nearby will be algorithms. May be
>> in Wolfram system 📯

>> Alex [2]
>>
>> [1]

>> [3].

>
> --
>
> ---
> Marco Neumann
>
> --
> All contributions to this forum are covered by an open-source license.
> For information about the wiki, the license, and how to subscribe or
> unsubscribe to the forum, see http://ontologforum.org/info
> ---
> You received this message because you are subscribed to the Google
> Groups "ontolog-forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ontolog-foru...@googlegroups.com.
> To view this discussion visit

> https://groups.google.com/d/msgid/ontolog-forum/CABWJn4QdhuSX4SNj%3DLS0s%2B7Uv8BBWT-hpNigA03v5ZvZZ0%2BLLw%40mail.gmail.com
> [4].

>
>
> Links:
> ------
> [1]
> https://docs.google.com/document/d/1N6vKV2FUly17U-gS7Sgzc67BxZUZz7h7mr6GJC9KexU/edit?tab=t.0
> [2] https://www.linkedin.com/in/ashkotin/
> [3]

> https://groups.google.com/d/msgid/ontolog-forum/CAFxxROQc86RoPg1YnY2xQheTyXuGiogsdiKsDDLQoKZeAODwHw%40mail.gmail.com?utm_medium=email&amp;utm_source=footer
> [4]
> https://groups.google.com/d/msgid/ontolog-forum/CABWJn4QdhuSX4SNj%3DLS0s%2B7Uv8BBWT-hpNigA03v5ZvZZ0%2BLLw%40mail.gmail.com?utm_medium=email&utm_source=footer

[John F Sowa]

Doug,

Your comments about going to the specialty ontologies for various application areas are very important.  Any big ontology that supports everything will require an open-ended collection of special subontologies (AKA branches at or near the top level of any general ontology}.  The only limit to the number of special cases is the amount of time, effort, and $$$ to develop them.

I did a bit of searching of the Cyc website.  They are still serving their traditional customers, and they have adopted and adapted LLMs for some purposes.  It would be interesting to get more info about how they relate their symbolic methods to the LLM methods.

This is a topic  that Arun Majumdar and I will be discussing in April, and it would be good if we could get a talk about how Cycorp is relating the two kinds of representations.

Lenat presented a good talk at the 2022 Ontology Summit, and he was scheduled to give another talk at the 2023 summit.  Unfortunately, he died.

John

 

---

From: do...@foxvog.org

[Marco Neumann]

Thank you, Alex, yes an apt observation. I would say that the editors tried to be a bit more pragmatic and RDF developer focused with the RDF Semantics 1.1 version. (2014)

And wordings like "interpretation", "entailment", "rule" and "possible world" are closely related to the extension of Tarski′s Semantics for RDF.  In this "world", truth is a correspondence between a sentence (SPO) and a model ("world"). For RDF, this is extended through a simple "Interpretation", which consists of: resources, properties, denotation and extension mapping. 

Marco 

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CAFxxROTQbd0otTKESvPRNzVfz0S16WQAzH5-kvYchb%3DKcLY%2Beg%40mail.gmail.com.

--

---
Marco Neumann

[Alex Shkotin]

Doug,

https://standards.clarin.eu/sis/views/view-spec.xq?id=SpecOpenCyc has https://cyc.com/platform/opencyc/ but it does not work.

Is this https://github.com/asanchez75/opencyc a good reference?

A system of interconnected terms with definitions is a backbone of any theory.

It is interesting that it has a value without axioms - the source of a theory.

Alex

пт, 13 мар. 2026 г. в 22:50, <do...@foxvog.org>:

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/9ff167f1ed979779fc9693a855409d9b%40foxvog.org.

[Alex Shkotin]

Marco,

I haven't looked into this, and of course, if there's an entailment, it's worth looking into.

I'll just note for now that when we study the language itself and its semantics, rather than the semantics of what can be written in it, it's worth first looking at its keywords:

https://www.w3.org/TR/rdf-syntax-grammar/#section-Namespace 

"The RDF namespace IRI (or namespace name) is http://www.w3.org/1999/02/22-rdf-syntax-ns# and is typically used in XML with the prefix rdf although other prefix strings may be used. The RDF Vocabulary is identified by this namespace name and consists of the following names only:

Syntax names — not concepts

RDF Description ID about parseType resource li nodeID datatype

Class names

Seq Bag Alt Statement Property XMLLiteral List

Property names

subject predicate object type value first rest _n
where n is a decimal integer greater than zero with no leading zeros.

Resource names

nil"

There's something to think about - very interesting keywords:

Seq Bag Alt Statement Property XMLLiteral List subject predicate object type value first rest _n nil

Alex

пт, 13 мар. 2026 г. в 23:42, Marco Neumann <marco....@gmail.com>:

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CABWJn4QLzyKoprNX5O2Tot8-NSp%3DyqibSkT4hYnVXf%2BJxJc5Ew%40mail.gmail.com.

[Ravi Sharma]

I plan to comment on topic soon but take this opportunity to thank all in this dialog

especially Amit Sheth as he assumes greater responsibilities relating to AI

regards

Thanks.

Ravi

(Dr. Ravi Sharma, Ph.D. USA)

NASA Apollo Achievement Award

​Former Scientific Secretary ISRO HQ

Ontolog Board of Trustees

Particle and Space Physics

Senior Enterprise Architect

SAE Fuel Cell Standards Member

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CAFxxROSSOT3CzRDx1YU5w23hbBaZF%3DTrfSZALmU5C5%3D7JazZuQ%40mail.gmail.com.

[Alex Shkotin]

Doug,

This distinction between terms and facts in DL-ontology is the distinction between a T-box and an A-box. As Gary mentioned.

Gary is silent, and I think we've discussed this topic enough.

🕉️

Alex

пт, 13 мар. 2026 г. в 23:26, <do...@foxvog.org>:

To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/9f7a7828b63817880ae62402cab35242%40foxvog.org.