Re: W3C Context Graph Kickoff!

Hi Juan, I agree it is exciting!

Learning lesson #1 for me: I had put details behind a link, but going
forward I'll put everything directly in the email. One less click to ask of
you.

For the kickoff I'll share a short video — a real example using SEC filings
that shows the problem we're solving, and a few open building blocks we're
making available to the community.

We'll cover short and long-term goals, cover what participation can look
like, what people want to work on, and what kinds of commitments make sense
for where they are.

Looking forward to meeting you all, thank you for joining us on this
adventure. ;)

*Calendar link and details below...*

W3C Context Graph Community Group — Email #001

Date: 2026-03-17
From: Ron Itelman, Community Group Chair
To: W3C Context Graph Community Group Members
Subject: Welcome — Committee Chairs, Kickoff Dates, and How We Work Together
------------------------------

Hi everyone,

I'm Ron Itelman, chair of the W3C Context Graph Community Group. I'm
grateful for the participation this group is already seeing, and I wanted
to share three things: a bit about where this work comes from, who's
leading the committees, and how I'd like us to work together.
Where This Started (A Short Origin Story)

I started this work doing R&D in banking around AI accuracy in finance. I
quickly ran into the roadblocks many people face daily:

   - I couldn't get the team owning the data catalog to give me access.
   - I had no authority to impose an ontology, and none existed that I was
   aware of.
   - All I could control was my own local system.

That constraint turned out to be the insight: start with what you have, not
what you want. I laid out the uncertainty — which columns were ambiguous,
which definitions were missing, which assumptions I was making — and
collapsed it piece by piece.

I saw the same pattern everywhere: Every team was doing context engineering
independently — constantly adapting prompts, running evals in their own
way, using LLMs to judge LLMs. It was uncertainty stacked on uncertainty. I
wanted to ground the effort in something that could actually be measured.

That led to a formal protocol for detecting and resolving misalignment at
system boundaries — not by imposing a framework, but by giving any system a
way to represent what it knows, what it doesn't know, and what it needs to
ask. That's what the context graph does, and that's what we're building
together.
Committee Chairs

Each committee covers a different part of the architecture. Please review
the full descriptions and decide which group (or groups) you'd like to
participate in:

*Each committee within the Context Graphs Community Group leads a specific
surface of this architecture.*
Committee Chair Scope
Coherence Protocol Ron Itelman Core protocol specification: four facets,
canonical claim form, Halt/Ask/Act, dependency ordering
Semantic Alignment Kurt Cagle Interface with existing semantic web
infrastructure (OWL, RDF, SHACL, SKOS)
Decision Interface Dr. Lorien Pratt Contract between coherence measurements
and decision models
Applied Knowledge Audrey Depeige Bridge between specification, deployment,
and measuring benefit
Syntax & Serialization Juan Cruz Viotti Wire formats, Intent Map
specification, canonical claim form serialization
Business & Industry Anthony Marquardt Use cases, pilots, and the business
case for adoption
Agentic AI *Pending* Agent-to-agent and agent-to-system boundary crossings
Error Analysis and Theory *Pending* Error classification, coherence
scoring, and measurement theory for the Halt/Ask/Act thresholds
Communications and Operations *Pending* Design simple ways to self-organize
effectively

Thank you to all the committee chairs for stepping up to lead this work.
I'll be honest — I'm amazed at the group we've assembled. Semantic web
architects, decision scientists, applied researchers, JSON/API engineers,
and business strategists all in one room. Every one of them is someone I
learn from, and the opportunity to work alongside them is not lost on me.

If you're interested in helping with group communications and coordination,
there's an open slot — reach out to me directly.

Full descriptions are here: Committee Chairs README
<https://github.com/W3C-Context-Graph-Community-Group/Charter>
How We Work Together

This group is community driven. The chairs bring a starting foundation — a
theoretical paper, a glossary, and an architectural framework. But the
specification will be shaped by what you bring to it. The specification
follows the problems, not the other way around.

If you see a problem we're not addressing, bring it. If you have boundary
problems from your own systems — the time your systems passed every check
and the answer was still wrong — those are exactly the use cases that will
make this standard real.
Kickoff Calls
<https://github.com/W3C-Context-Graph-Community-Group/Charter/blob/main/communications/Email%201%20-%20Kickoff.md#kickoff-calls>

I'd like to host two kickoff calls so people can choose which works better
for them, and if you can't make either time, please ping me and we'll
figure something out.

   -

   Option 1: March 24th
   Tuesday, March 24 · 12:00 – 1:00pm
   Time zone: America/Denver
   Video call link: https://meet.google.com/zmq-ztjj-qxx
   Or dial: ‪(US) +1 339-545-4323‬ PIN: ‪682 982 936‬#
   -

   Option 2: March 31st
   Tuesday, March 31 · 12:00 – 1:00pm
   Time zone: America/Denver
   Video call link: https://meet.google.com/bqr-nqsc-wut
   Or dial: ‪(US) +1 267-870-7492‬ PIN: ‪471 927 350‬#

Each call will cover the same content: brief introductions, a walk through
of the committees and how to get involved, and a chance to ask questions.
Attend whichever works for you.

Looking forward to meeting everyone.

Ron Itelman W3C Context Graph Community Group Chair



On Tue, Mar 17, 2026 at 2:45 PM Juan Cruz Viotti <jv@jviotti.com> wrote:

> Hey Ron,
>
> Exciting! Do you plan to send an invite or create a shared Google Calendar
> that we can subscribe to? Otherwise we might incorrectly book it!
>
> --
> Juan Cruz Viotti
> https://www.sourcemeta.com
>
>
>
>
>
>
>