15:56:36 <RRSAgent> RRSAgent has joined #rdf-star
15:56:40 <RRSAgent> logging to https://www.w3.org/2026/08/27-rdf-star-irc
15:56:41 <ktk> meeting: RDF-star WG meeting
15:56:48 <fsasaki> fsasaki has joined #rdf-star
15:56:50 <ktk> Agenda: https://www.w3.org/events/meetings/e5234c80-4c06-4c6b-af43-c78a1dbd390a/20260827T120000/
15:56:52 <agendabot> clear agenda
15:56:52 <agendabot> agenda+ Approval of minutes from the last two meetings: -> 1 https://www.w3.org/2026/08/13-rdf-star-minutes.html  -> 2 https://www.w3.org/2026/08/20-rdf-star-minutes.html
15:56:52 <agendabot> agenda+ CR Drafts
15:56:52 <agendabot> agenda+ Presentation of an LLM based 1.2 implementation (by lisp) -> 3 https://www.w3.org/2026/08/20-rdf-star-minutes.html#f513
15:56:53 <agendabot> agenda+ Review of open actions, available at -> 4 https://github.com/orgs/w3c/projects/20/views/3
15:56:55 <agendabot> agenda+ Review of open PRs, available at -> 5 https://github.com/orgs/w3c/projects/20/views/4
15:56:58 <agendabot> agenda+ Identifying issues to solve before CR -> 6 https://github.com/orgs/w3c/projects/20/views/5
15:57:01 <agendabot> agenda+ Any Other Business (AOB), time permitting
15:57:01 <ktk> RRSAgent, draft minutes
15:57:03 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html ktk
15:57:08 <ktk> RRSAgent, make log public
15:58:09 <ktk> regrets+ rubensworks
15:58:25 <tl> tl has joined #rdf-star
15:58:54 <AndyS> AndyS has joined #rdf-star
15:59:33 <AZ> AZ has joined #rdf-star
15:59:49 <AndyS> agenda?
16:00:44 <Dominik_T> Dominik_T has joined #rdf-star
16:01:02 <ora> present+
16:01:04 <ora> chair+
16:01:14 <fsasaki> scribe: fsasaki
16:01:29 <fsasaki> present+
16:01:29 <doerthe> doerthe has joined #rdf-star
16:01:30 <Dominik_T> present+
16:01:40 <doerthe> present+
16:01:48 <gtw> present+
16:01:50 <AZ> present+
16:01:57 <AndyS> present+
16:02:01 <lisp> lisp has joined #rdf-star
16:02:04 <olaf> olaf has joined #rdf-star
16:02:06 <tl> present+
16:02:10 <lisp> present+
16:02:10 <olaf> present+
16:02:22 <pchampin> present+
16:03:06 <TallTed> present+
16:03:20 <pfps> present+
16:03:44 <ora> Zakim, open item 1
16:03:44 <Zakim> agendum 1 -- Approval of minutes from the last two meetings: -> 1 https://www.w3.org/2026/08/13-rdf-star-minutes.html  -> 2 https://www.w3.org/2026/08/20-rdf-star-minutes.html --
16:03:47 <Zakim> ... taken up [from agendabot]
16:03:52 <Tpt> present+
16:03:54 <Tpt> scribe+
16:03:55 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html TallTed
16:04:14 <niklasl> niklasl has joined #rdf-star
16:04:38 <Tpt> scribe-
16:05:01 <ora> PROPOSAL: Accept last two week's minutes
16:05:18 <pfps> +1
16:05:19 <ora> +1
16:05:22 <fsasaki> ted: minor type on the 20th, it is minor, can be ignored
16:05:24 <Dominik_T> +0 (not present)
16:05:28 <niklasl> +1
16:05:29 <olaf> +1
16:05:31 <fsasaki> +1
16:05:34 <AndyS> +1
16:05:39 <pchampin> +1
16:05:42 <Souri> Souri has joined #rdf-star
16:05:42 <gtw> +1
16:05:45 <pfps> s/type/typo/
16:05:50 <lisp> +1
16:05:50 <Souri> present+
16:05:58 <doerthe> +1
16:06:00 <ktk> +1
16:06:03 <Souri> +1
16:06:04 <tl> +1
16:06:10 <AZ> +0  (not present at either meeting)
16:06:19 <ora> RESOLVED: Accept last two week's minutes
16:06:26 <TallTed> s/ted: minor/TallTed: minor/
16:06:30 <ora> Zakim, open next item
16:06:30 <Zakim> agendum 2 -- CR Drafts -- taken up [from agendabot]
16:06:31 <TallTed> +1
16:06:46 <Dominik_T> RRSAgent, draft minutes
16:06:47 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html Dominik_T
16:07:19 <fsasaki> pchampin: recent version of w3c process extended notion of CR
16:07:26 <fsasaki> ... we now have CR snapshots
16:07:33 <fsasaki> ... and CR drafts
16:07:41 <fsasaki> ... we have not used the latter notion yet
16:08:02 <AndyS> q+ to ask what the difference is.
16:08:03 <fsasaki> ... we are frozen in the three specs in the state they were
16:08:15 <fsasaki> ... we could have enabled automated publication syncing all changes into CR drafts
16:08:42 <fsasaki> ... question: do we want to do that, as long as we do not make substantive changes it does not preclude us to got to REC
16:08:52 <fsasaki> ... in case we make substantive changes we need a new snapshot
16:09:07 <fsasaki> ... so question to the group: do we want to re-enable automatic publication
16:09:20 <fsasaki> ... to give more visibility to changes we made since CR
16:09:55 <fsasaki> ... e.g. RDF concepts, the CR was published april 7th, we did many adjustments
16:10:08 <fsasaki> ... all these are only visible if you dig deeper
16:10:19 <fsasaki> ... it is mainly a question of visibility
16:10:34 <ora> ack AndyS
16:10:34 <Zakim> AndyS, you wanted to ask what the difference is.
16:10:35 <fsasaki> ... we need to work with CR snapshots anyway for moving to REC
16:10:53 <fsasaki> andy: a question, what is the difference between CR snapshot and CR draft
16:10:55 <Tpt> -1
16:11:00 <fsasaki> ... what goes to CR: snapshot or draft?
16:11:25 <fsasaki> pchampin: many groups are now publishing on TR what they have as an editors draft
16:11:39 <fsasaki> ... they want to have the TR version as recent as possible
16:11:51 <fsasaki> ... CR draft is any version after a CR snapshot was published
16:11:59 <fsasaki> ... we are now in CR, calling for implementations
16:12:07 <fsasaki> ... and we continue to improve the document
16:12:33 <fsasaki> ... if we want to say "there have been substantive changes", we can move from a CR draft to a new CR snapshot
16:12:45 <j22> j22 has joined #rdf-star
16:12:51 <fsasaki> ... if there are no substantive hcanges, we can go from the latest draft to RC
16:12:51 <j22> present+
16:12:55 <fsasaki> S/RC/REC/
16:13:17 <fsasaki> andy: suggest to have the live version as recent as possible
16:13:26 <fsasaki> ora: what is the downside? is there any?
16:13:50 <fsasaki> pchampin: the same as having ecidna published
16:14:12 <fsasaki> ... some WG want to have publications more meaningful to express WG endorsement
16:14:25 <fsasaki> ... for this they do not sync ecidna with a separate branch
16:14:36 <fsasaki> ... they merge then only once things are ready to go to TR
16:14:48 <fsasaki> ... we have adopted the "release early, release often" model during draft state
16:14:55 <fsasaki> ... we may consider that CR is different
16:15:07 <fsasaki> ... or remain in the "release early, release often" mode
16:15:15 <fsasaki> andy: so no difference towards the timing
16:15:18 <fsasaki> pchampin: yes
16:15:18 <ora> PROPOSAL: Move to the new CR snapshot/draft model?
16:15:21 <ora> +1
16:15:21 <niklasl> +1
16:15:25 <fsasaki> +1
16:15:26 <pchampin> +1
16:15:27 <olaf> +1
16:15:28 <Dominik_T> +1
16:15:28 <AZ> +1
16:15:29 <lisp> +1
16:15:30 <AndyS> +1
16:15:31 <doerthe> +1
16:15:31 <gtw> +1
16:15:32 <tl> +1
16:15:34 <Souri> +1 to CR draft as long as it is indicated prominently that it is indeed a draft and has a pointer to the official CR-snapshot
16:16:37 <fsasaki> pchampin: CR drafts are drafts, now checking if they point to the CR snapshot
16:17:01 <fsasaki> ... the history link shows the latest snapshot
16:17:04 <pchampin> https://www.w3.org/standards/history/vc-data-model-2.0/
16:17:40 <fsasaki> pchampin: that is the histroy. You see always the link to the history in a draft
16:18:52 <fsasaki> ora: so you want to have one click?
16:19:06 <fsasaki> souri: if there is a search involved it would not be obvious
16:19:16 <fsasaki> ora: other thoughts?
16:19:47 <fsasaki> ... should we should propose this to the people who work on the publishing tooling?
16:19:57 <fsasaki> andy: we can use the "latest published version" link
16:20:06 <pchampin> q+
16:20:16 <ora> ack pchampin
16:20:22 <fsasaki> pchampin: "latest published version" is most recent version on tr
16:20:31 <fsasaki> ... if that is the cr draft, you will end up with that
16:20:40 <fsasaki> .. some specs have a latest rec, that is it
16:21:21 <fsasaki> pchampin: we can create custom headers and try to maintain one to the latest snapshot
16:21:26 <fsasaki> souri: would be great
16:21:32 <ora> RESOLVED: Move to the new CR snapshot/draft model?
16:21:47 <fsasaki> ora: pchampin will set things up
16:21:50 <ora> Zakim, open next item
16:21:50 <Zakim> agendum 3 -- Presentation of an LLM based 1.2 implementation (by lisp) -> 3 https://www.w3.org/2026/08/20-rdf-star-minutes.html#f513 -- taken up [from agendabot]
16:21:56 <Dominik_T> RRSAgent, draft minutes
16:21:57 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html Dominik_T
16:22:09 <lisp> https://dydra.com/data/protocol-process-1.pdf
16:22:27 <lisp> https://dydra.com/data/protocol-process-2.pdf
16:22:38 <lisp> https://dydra.com/data/1_2-process.pdf
16:23:04 <fsasaki> lisp: above links are three diagrams about using an LLM with W3C docs to perform development tasks
16:23:29 <fsasaki> ... was an exchange in sparql task force regarding the state of the sparql test suite
16:23:48 <fsasaki> ... we saw that specs where not in sync, I wanted to check how an LLM can help
16:23:57 <fsasaki> ... that was first step, see the first diagram
16:24:18 <fsasaki> ... in all cases the prompts where very simple. I asked to make corrections, deletions
16:24:39 <fsasaki> ... the LLM did that and produced a version that gregory described as deficient in some ways
16:24:51 <fsasaki> ... so I looked into how this can be made better
16:25:08 <fsasaki> ... the 2nd diagram provides more control, more constraints on the LLM
16:25:33 <fsasaki> ... to provide controls
16:26:28 <fsasaki> ... so that links between assets could be checked better
16:26:56 <fsasaki> ora: what is the difference between solid, dashed and dottet lines
16:27:05 <fsasaki> lisp: solid is data flows from documents
16:27:17 <fsasaki> ... solid lines forward are references forward from the LLM
16:27:34 <fsasaki> ... the dashed lines were prompts from me
16:27:58 <fsasaki> ... then setting up a test runner to run LLM until agreement is done, with a semi automated loop
16:28:16 <fsasaki> ... the pattern got applied
16:28:31 <fsasaki> ... to the process I followed over 6 weeks to implement sparql 1.2
16:28:38 <fsasaki> .. from the start of doing turtle
16:28:56 <fsasaki> ... to doing n-quads, n-triples and sparql-protocol and graph protocol
16:29:15 <fsasaki> ... pointing an llm to the rec, concepts, and asking to write a requirements document for the implementation
16:29:37 <fsasaki> ... the requirements document was about what it had to do about the implicit aspects
16:29:57 <fsasaki> ... then I asked the llm to re-write the implementation, again in an automated loop
16:30:24 <fsasaki> ... the execution in testing also for triple terms was with very simple prompting
16:30:56 <fsasaki> TallTed: did the requirements document have MUST and SHOULD etc.
16:31:26 <fsasaki> lisp: the doc had also SHOULD etc., the implementation was only for MUST
16:32:02 <fsasaki> ora: so an LLM wrote an implementation?
16:32:22 <fsasaki> lisp: the LLM took an conformant implementation for 1.1 so that it passes all tests for 1.2.
16:33:28 <fsasaki> lisp: lisp is not creating an obstacle for the LLMs
16:33:48 <ora> s/lisp is not/Common Lisp is not/
16:33:50 <fsasaki> ... if one wants to look at loosing an LLM for providing support for implementation - this was six weeks of work
16:34:23 <fsasaki> ... it was able to extract the requirements from the implementation, modifying the existing implementation
16:35:07 <fsasaki> ... and some LLM tooling has detailed logs of what was done
16:35:28 <fsasaki> ... recorded in very explicit terms
16:35:37 <pchampin> q+
16:35:42 <ora> ack pchampin
16:36:38 <fsasaki> pchampin: the idea that llm tooling gives a detailed account of what was done
16:36:58 <fsasaki> ... the idea is to say: someone takes responsibiilty of what was done
16:37:20 <lisp> q+
16:37:28 <ora> ack lisp
16:37:42 <fsasaki> pchampin: not sure if LLM tooling responds to that concern
16:37:43 <Souri> s/llm/LLM/
16:38:05 <fsasaki> lisp: the LLM has a certain ability to interpret an LLM
16:38:09 <pfps> q+
16:38:17 <ora> ack pfps
16:38:36 <fsasaki> pfps: pchampin point is:
16:38:47 <lisp> q+
16:38:55 <fsasaki> ... it does not help to produce a large document that makes explicit from which comes from
16:39:06 <fsasaki> ... that can not be done by a WG
16:39:24 <fsasaki> ... just using a long document does not help as a justification in a WG
16:39:25 <ora> ack lisp
16:39:41 <fsasaki> lisp: very likely one would not read that
16:39:41 <TallTed> q+
16:39:46 <ora> ack TallTed
16:39:48 <fsasaki> ... but have an LLM that can use it
16:40:04 <fsasaki> TallTed: LLMs still concat their own information
16:40:20 <Souri> s/concat/concoct/
16:40:41 <fsasaki> ... I recently tried to process pronounciation rules
16:41:11 <fsasaki> ... the LLM came back with full assurance about some rules but was not correct
16:41:14 <lisp> q+
16:41:18 <tl> q+
16:41:27 <fsasaki> TallTed: I had to say "this is wrong"
16:41:46 <fsasaki> ... the meat of the W3C policy is: a human has to have reviewed what came forward
16:42:17 <fsasaki> TallTed: I need a human reviewing it
16:43:04 <ora> ack lisp
16:43:30 <fsasaki> lisp: that is why I went through the history to provide a perspective of the problem
16:43:38 <fsasaki> ... the first pass was unguided
16:43:44 <ora> q+
16:43:51 <fsasaki> ... the 2nd pass had a requirements doc which I reviewed
16:44:06 <fsasaki> ... I reviewed the transcript as it worked
16:44:16 <fsasaki> ... the full process implementing 1.2. had the same order of control
16:44:27 <fsasaki> ... it came up with proposals
16:44:41 <fsasaki> ... where I had to go back to ask: where did you get this from?
16:44:55 <fsasaki> ... it was given a process which had controls on it
16:45:01 <fsasaki> ... it had at the end the tests
16:45:03 <ora> q-
16:45:40 <fsasaki> ... it executed all tests except the ones I said: you can skip
16:45:55 <fsasaki> ... so if these controls are considered: the concerns are addressable
16:46:09 <fsasaki> ora: now handing over chairing to pchampin, need to drop out
16:46:25 <pchampin> ack tl
16:46:35 <pchampin> zakim, close queue
16:46:35 <Zakim> ok, pchampin, the speaker queue is closed
16:46:37 <fsasaki> tl: the work seems to rely on test cases a lot
16:46:44 <fsasaki> ... not everything is covered in test cases
16:46:57 <fsasaki> ... how do you or does the LLM deal with that?
16:47:19 <fsasaki> lisp: the success criteria was with regards to the W3C test suite
16:47:30 <pchampin> Zakim, open next item
16:47:30 <Zakim> agendum 4 -- Review of open actions, available at -> 4 https://github.com/orgs/w3c/projects/20/views/3 -- taken up [from agendabot]
16:48:08 <fsasaki> niklasl: worked on my action item
16:48:15 <pchampin> zakim, open next item
16:48:15 <Zakim> agendum 4 was just opened, pchampin
16:48:25 <pchampin> zakim, close item 4
16:48:25 <Zakim> agendum 4, Review of open actions, available at -> 4 https://github.com/orgs/w3c/projects/20/views/3, closed
16:48:27 <pchampin> zakim, open next item
16:48:28 <Zakim> I see 3 items remaining on the agenda; the next one is
16:48:28 <Zakim> 5. Review of open PRs, available at -> 5 https://github.com/orgs/w3c/projects/20/views/4 [from agendabot]
16:48:28 <Zakim> agendum 5 -- Review of open PRs, available at -> 5 https://github.com/orgs/w3c/projects/20/views/4 -- taken up [from agendabot]
16:48:59 <fsasaki> pchampin: anything to discuss?
16:49:17 <pchampin> subtopic: https://github.com/w3c/rdf-concepts/pull/282
16:49:17 <gb> s|https://github.com/w3c/rdf-concepts/pull/282|-> PR 282 Reference RFC 9839 in Security Considerations (by domel) https://github.com/w3c/rdf-concepts/pull/282
16:49:41 <fsasaki> andy: 282 is ready
16:49:56 <fsasaki> ... on sparql query, 416 is ready
16:50:02 <fsasaki> .. and 400 too
16:50:15 <olaf> q+ no, 400 is not
16:50:21 <olaf> q+
16:50:47 <fsasaki> pchampin: there is a change request from olaf on 400
16:51:06 <olaf> q-
16:51:13 <fsasaki> andy: on turtle, 157 is ready
16:52:28 <TallTed> s|157|https://github.com/w3c/rdf-turtle/pull/157|
16:52:55 <fsasaki> .. on rdf tests, looking now
16:53:31 <fsasaki> pchampin: so 416 is ready on sparql
16:53:32 <TallTed> s/and 400 too|and https://github.com/w3c/sparql-query/pull/400 too|
16:53:39 <fsasaki> andy: will look and do that
16:53:42 <pchampin> q?
16:53:51 <lisp> q+
16:54:14 <pchampin> https://github.com/w3c/rdf-n-quads/pull/108
16:54:15 <gb> https://github.com/w3c/rdf-n-quads/pull/108 -> PR 108 Undefined errors (by lisp)
16:54:21 <fsasaki> pchmapin: james pr on n-quads was discussed previously
16:54:22 <TallTed> s|416 is ready|https://github.com/w3c/sparql-query/pull/416 is ready|
16:54:31 <fsasaki> ... the essence of my comment on that pr is:
16:54:34 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html TallTed
16:54:52 <fsasaki> ... once we propagate change that have been made in n-triples and then this to n-triples then conflicts will arise
16:55:10 <fsasaki> ... so would suggest to make the pr against n-quads
16:55:25 <fsasaki> ... if we do that changes we have a conflict later down the road
16:55:43 <pchampin> q?
16:55:52 <fsasaki> lisp: I saw your mail, the group should decide on that
16:55:52 <pchampin> ack lisp
16:56:04 <fsasaki> ... you said that some aspects in n-triples should be retained
16:56:13 <fsasaki> ... I do not think that is true
16:56:20 <fsasaki> ... the group should decide on that
16:56:37 <fsasaki> pchampin: we have a chicken and egg problem
16:56:42 <fsasaki> ... any PR is just a proposal
16:56:51 <fsasaki> ... the current PR does not show what will be removed
16:57:17 <fsasaki> lisp: n-triples and n-quads should agree by including the n-quads content
16:57:21 <fsasaki> .... that is what I will do
16:57:52 <fsasaki> pchampin: we talk past each other. I would like to keep the n-quads PR in a parking lot for the moment
16:58:01 <fsasaki> ... and make a PR on n-triples first
16:58:22 <fsasaki> ... the most up to date doc is n-triples
16:58:50 <fsasaki> ... so make an PR in n-triples that is equivalent, the group can decide on that
16:59:11 <fsasaki> lisp: OK, will do
16:59:34 <pchampin> q?
16:59:41 <fsasaki> pchampin: any other business?
16:59:52 <pchampin> RRSAgent, make minutes
16:59:53 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html pchampin
16:59:56 <fsasaki> adjourned
17:01:03 <pchampin> s/pchmapin/pchampin
17:02:22 <pchampin> present+ niklasl
17:02:35 <pchampin> s/andy:/AndyS:/g
17:02:40 <pchampin> RRSAgent, make minutes
17:02:42 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html pchampin
17:03:41 <pchampin> s|and 400 too|and https://github.com/w3c/sparql-query/pull/400 too|
17:03:44 <pchampin> RRSAgent, make minutes
17:03:45 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html pchampin
17:05:21 <pchampin> s|282 is ready|https://github.com/w3c/rdf-concepts/pull/283 is ready
17:06:09 <pchampin> s|416 is ready|https://github.com/w3c/sparql-query/pull/416 is ready
17:06:10 <pchampin> RRSAgent, make minutes
17:06:12 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html pchampin
17:07:10 <pchampin> s|so https://github.com/w3c/sparql-query/pull/https://github.com/w3c/sparql-query/pull/416|so https://github.com/w3c/sparql-query/pull/416
17:07:34 <pchampin> s|query, 416 is ready|query, https://github.com/w3c/sparql-query/pull/416 is ready
17:07:36 <pchampin> RRSAgent, make minutes
17:07:37 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html pchampin
17:08:28 <pchampin> s|subtopic: -> PR 282 Reference RFC 9839 in Security Considerations (by domel) https://github.com/w3c/rdf-concepts/pull/282|
17:08:30 <pchampin> RRSAgent, make minutes
17:08:32 <RRSAgent> I have made the request to generate https://www.w3.org/2026/08/27-rdf-star-minutes.html pchampin
17:25:08 <denkeni> denkeni has joined #rdf-star