- From: Neil Soiffer <soiffer@alum.mit.edu>
- Date: Wed, 6 Nov 2024 17:13:42 -0800
- To: "www-math@w3.org" <www-math@w3.org>
- Message-ID: <CAESRWkDLDmhocoDut0pumvm_74DMzX6MrAN-USzaPAZuVRv9tw@mail.gmail.com>
Attendees: - Neil Soiffer - Louis Maher - Bruce Miller - David Farmer - Murray Sargent - Bert Bos - Paul Libbrecht - David Carlisle - Moritz Schubotz <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0-regrets> Regrets - Deyan Ginev - Paul Libbrecht - Patrick Ion <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0-action-items>Action Items North America moves its clocks back (for the places that observed daylight savings time) this coming Sunday (November 3). So, the meeting will start at its traditional time for individuals in Europe. *CONSENSUS* Currently, for the full spec, BB has arranged to publish a new WD on the /TR page every time an update to the editors' draft is pushed to GitHub. The group agrees that we should do this for the core spec also. DC will make this change. *CONSENSUS* DC has an XML entities document which we agree to have automatically moved into TR space. <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0-2-mathml-4-issues->2. MathML 4 issues: <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0--a-href-https-github-com-w3c-mathml-pull-511-discussion-about-update-for-describing-matching-concepts-a->Discussion about update for describing matching concepts <https://github.com/w3c/mathml/pull/511> <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0--a-href-https-brucemiller-github-io-mathml-spec-html-mixing_intent-bruce-39-s-suggested-rewrite-a->Bruce's suggested rewrite <https://brucemiller.github.io/mathml/spec.html#mixing_intent> <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0-deyan-39-s-suggestions-for-changes-to-bruce-39-s-rewrite-a-href-https-github-com-brucemiller-mathml-pull-1-small-number-of-diffs-a->Deyan's suggestions for changes to Bruce's rewrite (small number of diffs) <https://github.com/brucemiller/mathml/pull/1> *ACTION* NS: We should add the literal property to section 5.3. *ACTION* NS: We need to make a note saying that: When we mention 'dictionary' we say that "dictionary" is a convenient metaphor for what we are doing. We may use a dictionary, or an AI based solution. A dictionary is a mapping of concepts to speech items. *ACTION* NS: So instead of a core concept list, we should call it a core concept dictionary. A dictionary does not necessarily indicate that the items are ordered. It could just be a method to map concepts to speech items. *ACTION* NS found some mistakes and will send them to BM. BM will then make a PR and merge his branch. *CONSENSUS* Change list to dictionary. <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0-agenda> Agenda <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-0-1-announcements-updates-progress-reports>1. Announcements/Updates/Progress reports North America moves its clocks back (for the places that observed daylight savings time) this coming Sunday (November 3). So, the meeting will start at its traditional time for individuals in Europe. *CONSENSUS* Currently, for the full spec, BB has arranged to publish a new WD on the /TR page every time an update to the editors' draft is pushed to GitHub. The group agrees that we should do this for the core spec also. DC will make this change. *CONSENSUS* DC has an XML entities document which we agree to have automatically moved into TR space. DC: There has been a lot of work to make it possible to use MathML inside pdf documents. DC: In Pdf 2.0, you can tag mathematics with MathML, either as an associated file or internally. It's sort of in line within the pdf stream for MathML tags. DC: The automatic generation of MathML by LuaLaTeX is quite new in the standard release. MoS: Is there a place where I can get the official version of the latest Xml representation because that's much easier to consume with the script than the HTML? DC: Just ping me offline if you cannot find it and I will give you the GitHub links. MoS: So, when I want to process some of the contents of the standards, I should just go to GitHub and get it from there. DC: The sources are Respect sources so they're HTML. <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-1-2-mathml-4-issues->2. MathML 4 issues: <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-1--a-href-https-github-com-w3c-mathml-pull-511-discussion-about-update-for-describing-matching-concepts-a->Discussion about update for describing matching concepts <https://github.com/w3c/mathml/pull/511> Important: Bruce did a rewrite <https://brucemiller.github.io/mathml/spec.html#mixing_intent> of some of the intent sections that we will discuss. It is too long to read at the meeting. It will probably take 5-10 minutes to read on your own, depending on how closely you read it. Please take a few minutes before the meeting to read it so we don't have to spend time reading it at the meeting <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-1--a-href-https-brucemiller-github-io-mathml-spec-html-mixing_intent-bruce-39-s-suggested-rewrite-a->Bruce's suggested rewrite <https://brucemiller.github.io/mathml/spec.html#mixing_intent> NS: The rewrite was good. NS: Did the intro change much? BM: no. BM: For the most part I was kind of collecting up things that seemed like they were getting discussed a little too early, or they were getting discussed in several places. NS: I don't think you changed anything in the grammar, except for the concept or literal discussion. Is that correct? BM: I pulled some stuff out of there from the concept and literal section and moved them down to the next section, otherwise the grammar is pretty much the same. I hoped that I was not changing anything, just rearranging. NS: You started making major changes in section 5.2 "Intent Concept Dictionaries". BM: This section discussed implicitly something, a dictionary, a list, or whatnot. It should contain the concepts from core, and may include those from open, but neither of those are absolutely required. BM: Whatever it is, it acts like a dictionary. NS: DG feels that AI is cut out of this definition of a dictionary. Have we prevented an AI solution? NS: We will discuss that later and go on with what you have changed. BM: 5.3 "intent properties". I did not make much change here. Perhaps it's too rich in examples. *ACTION* NS: We should add the literal property to section 5.3. 5.4 Using Intent Concepts and Properties: NS to BM, so this was your biggest area of focus? BM: Yes, this is where I tried to collect up all the pieces that was from the other places, and put them in one place together and go through it more consistently. BM: Are there arguments? Are there properties about the fixity? BM: I've added a bit about the case where there's no explicit arguments BM: If you have a concept, fixity, and arity, then it's a known concept. BM: If concept, fixity, and arity match, then you have a known concept. If they do not match, you have an unknown concept, so you would treat it as a literal. BM: If there's no concept component, if it's just a property or arguments, then you mostly treat it as a literal, but possibly modified by whatever the property asked for. BM: I did fold in the section about standalone properties. BM: I don't think that changed anything after this. <https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-5-migration2#cp-md-1-deyan-39-s-suggestions-for-changes-to-bruce-39-s-rewrite-a-href-https-github-com-brucemiller-mathml-pull-1-small-number-of-diffs-a->Deyan's suggestions for changes to Bruce's rewrite (small number of diffs) <https://github.com/brucemiller/mathml/pull/1> NS: DG dictionary lookup forces it towards an implementation, and this blocks an AI solution. DC: DG has introduced the concept of a match maid by AI and not dependent on ordered lists. DC: Main issue,. *ACTION* NS: We need to make a note saying that: When we mention 'dictionary' we say that "dictionary" is a convenient metaphor for what we are doing. We may use a dictionary, or an AI based solution. A dictionary is a mapping of concepts to speech items NS: I have some text that I can send to you (DC) that I had sent to DG, which was not totally spec worthy text. But basically, it says it uses the word metaphor, and says that as long as any solution conforms to making use of the concept name, and the properties, and number of arguments, then it's fine. NS: The AT would be trained on the dictionary, but be in a neural net. NS wanted to ask the groups opinion of matching that could be done by AI. MoS: The core and open lists should be merged together. MuS: I had a differential d and somehow I needed a way to say that, yes, the user really wants this To be pronounced a certain way versus, okay, this is a good idea, you know And I didn't see any way in our syntax to distinguish between those 2 cases. NS: We discussed having a terse or verbose reading mode where you're now making it clear that here's 2 different ways of saying it, and I'll leave it up to the AT to decide. DC: You decide which reading mode you want at the time of reading. NS: It sounds like everybody's more or less on board with what's going on. Modulo. Murray's comments about alternative readings. DC: We introduced the word list in many places, we should revisit the use of the word list. We discussed the differences between the words "list" and "dictionary". *ACTION* NS: So instead of a core concept list, we should call it a core concept dictionary. A dictionary does not necessarily indicate that the items are ordered. It could just be a method to map concepts to speech items. *ACTION* NS found some mistakes and will send them to BM. BM will then make a PR and merge his branch. *CONSENSUS* Change list to dictionary. [PI: suggestion for quibble:: as "three quarters" or "three over x" or may vocalize an inline fraction marked up as / => as "three quarters", in a specific case, or, more generally, "three over x" or may vocalize an inline fraction marked up with /
Received on Thursday, 7 November 2024 01:14:00 UTC