Re: Reification

Daniel,
All,

Thank you for those K3D hyperlinks.

Enabling articles to be about reified statements, e.g., whether an Action adheres to a Rule, could be part of a larger set of features to enable reasoning and discussion about both proposed and performed actions, e.g., environmental impacts.

While an article about "R(subj, obj)" could also be about "subj" and "obj" or, instead, about "subj" while just mentioning "obj", uses of reification would enable benefits including new precision with respect to the indexing of, searching for, and retrieval of articles.

As for how the use of reification under discussion might become a first-class pattern in use on the Web, there might be some user-experience design topics to consider to enable end-users to be able to easily and intuitively enter these metadata into their content-management systems when editing or publishing articles.

It seems that it would be easier to provide end-users with a URL-entry and thing-selection process for indicating that their article was about one or more things rather than to enable them to enter that their articles were about one or more statements about things.

That is, when making use of a Web-based GUI, e.g., in a CMS, to enter what their article was about, an end-user could paste a URL into an text-input box to get a selection process with respect to those things at or having that URL, e.g., to confirm that they wanted to select an Action that was described in embedded schema in a Web document at that URL (https://example.com/actions/123.html).

Similarly, an end-user could paste a second URL into a GUI widget to express that they intended to also enter a Rule available at that second URL (https://example.com/policy/rule-1.html).

It would take some doing to enable end-users to be able to easily and intuitively express that an article was about whether a selected Action adhered to a selected Rule.

Brainstorming, perhaps end-users could press a button, "form relation between things", select the subject thing, then the object thing, and then be provided an incremental search box or scrollable item-selection component to be able to select a relationship (from one or more configured ontologies) to relate the selected things. Then, after such a process, there would be a visual representation on-screen of the statement as a thing for them to enter that their article was about.

In my opinion, benefits, e.g., with respect to delivered search and Q&A capabilities, should outweigh costs, e.g., with respect to UX design, software development, end-user training, and, eventually, end-user effort (those extra gestures to form a statement relating things).

Also, even with the capability to express and to enter that an article was about "R(subj, obj)", not all end-users would consistently enter the best metadata for their articles into their CMS software...

So, perhaps, then, what end-users would want would be AI-enhanced tools or services to recommend complex metadata for their articles during their editing or publishing processes. In this way, end-users would be able to receive all of the benefits without as much, or any, of the costs.


Best regards,
Adam

P.S.: With respect to plausible knowledge notation (PKN), I found: https://w3c.github.io/cogai/pkn.html#reasoning-graphs .

________________________________
From: Daniel Ramos <capitain_jack@yahoo.com>
Sent: Tuesday, November 18, 2025 1:05 PM
To: public-aikr@w3.org <public-aikr@w3.org>; semantic-web@w3.org <semantic-web@w3.org>; Dave Raggett <dsr@w3.org>
Subject: Re: Reification


Adam, semantic‑web and AI‑KR folks,

Thank you for sharing the schema.org discussion and the JSON‑LD‑star example for “article about whether an Action adheresTo Rules.” That kind of reified statement is exactly the sort of structure I want to track explicitly in my work on spatial KR.

Concretely, in K3D we’ve been treating:

Actions, rules and statements as Nodes in a 3D House/Galaxy;
Relations like adheresTo / violates / discusses as rays and edges;
and time/adequacy as fields attached to those structures (e.g., when, by whom, with what confidence).
On the implementation side, we’re currently experimenting with a small extension to our GPU‑native RPN engine to support three‑valued logic and balanced ternary signals on PTX, e.g.:

adheres / violates / unknown,
consistent / contradictory / undecided,
attract / neutral / repel in semantic fields.
Those ternary values are intended to complement, not replace, richer KR formalisms: they sit alongside embeddings and symbolic structures and can be computed very efficiently on GPU. I see a lot of potential in combining:

patterns like your JSON‑LD‑star reification (Action + Rule + about relationships),
plausibility layers like Dave’s Plausible Knowledge Notation (PKN),
and these lightweight ternary fields for “how well does this action adhere to this rule, right now, in this domain?”
There is also a resource‑ and carbon‑angle here that Milton has been highlighting from the AI‑for‑Good / ICT4D side. Our design is explicitly aimed at:

small, recursive models rather than ever‑growing LLMs;
procedural compression and GPU‑local memory instead of giant host‑RAM arrays;
running serious KR + reasoning on mid‑range hardware rather than hyperscale datacenters.
If we want reified KR about actions, rules and impacts to be used in SIDS, indigenous communities, and low‑resource contexts, this question of representation + efficiency becomes very concrete.

If it’s helpful for the discussion, here are two implementation‑oriented notes from my side (not standards proposals, just working docs):

Spatial encoding of domains, concepts and relations (stars, rays, Garden, Museum):

https://github.com/danielcamposramos/Knowledge3D/blob/main/docs/SPATIAL_KR_VISUAL_ENCODING.md


Chain file for exploring ternary RPN semantics over PTX (balanced ternary / three‑valued logic):

https://github.com/danielcamposramos/Knowledge3D/blob/main/docs/RPN_TERNARY_SETUN_CHAIN.md


Projection of the carbon print impact in 10 years (If the industry adopts K3D):

https://github.com/danielcamposramos/Knowledge3D/blob/main/docs/CARBON_BLUEPRINT_10_YEAR_PROJECTION.md




I’d be very interested in seeing how a pattern like the one you showed (Action adheresTo Rule, reified and “about”‑ed by articles) could be treated as a first‑class KR pattern (RDF/JSON‑LD‑star + PKN‑style plausibility) and also as a spatial substrate that AIs and humans can actually navigate, introspect, and update over time.

Best regards,
Daniel