<tzviya> Meeting Chemistry Publishing

<scribe> scribenick: franco

<scribe> scribe: franco

<scribe> ACTION: item to george to learn zakim

george: approve minutes from may 21
... minutes approved
... have been sending emails to other email lists. the only place i need to send the email to is the chemistry mailing list. does that sound right?

dan: i was in there yesterday and i see only six people have joined the community group.

neil: there are 14 people in the group

kerry: as we think of more people, maybe we can reach out individually. if the group gets too
... big i dont know if itll ever get anything done

george: when we get the minutes i'll send them out only to this group and ...
... future invitations will only be on this email list. we may want to post ...
... something in the future to let people know that the chemistry group is here
... debrief on the join mathml chemistry call

kerry: dan and i were both on the call. dan, if you want to start off

dan: what we came away with was we can put something into mathml that will distinguish...
... when it reaches something that is chemistry notation. it could folow the rules that we
... outline for it
... and i think that was the main takeaway from it

kerry: we did get into some specific examples of what i consider to be ambiguities in the
... chemistry context. for example: K can represent different things. with the screen reader
... it typically just says k. i found this when i was lecturing. there are several other examples like
... with t, r, etc. we got into a little discussion about that and how to address those

dan: something about mapping certain characters into mathml and see how they work

george: is this a screen reader issue or a mathml problem. if mathml does differentiate
... between the different k's but the screen reader doesnt read it, then thats a screen reader
... problem

bill: id be interested to see what benetech's mathml cloud does. can it tell the difference?

neil: it is different in mathml. that is in the mathml. you can't necessarily blame the screen reader
... its whatever service they are using. they may be using different ways to convert it
... it could be the screen reader settings.
... so its hard to pinpoint the problem.
... if there is a difference , there are two possibilities: translation is not great, or the user settings are not set correctly

kerry: can you elaborate?

neil: if it's roman, it's roman; if its lowercase, its lowercase. mathplayer used to say cap k, and then it let it pass through to screen reader
... most screen readers will distinugish with a pitch change
... if you dont do that, maybe you wont hear a pitch change. the difference between the script and the roman, that could be translator itself taht converts the mathml into speech
... not saying roman k

kerry: i dont know how to make a script k in word, for instance

neil: its possible by default a variable will be done with a script
... an m that happens to represents a meter rather than a variable m.

kerry: how do we address these issues in mathml ?

neil: i dont think in the call any conclusions were made. it was just informative.
... we keep in mind as we go the issues that chemistry has. there may be a role = chemistry and readers just pick that up
... its possible you have to go down to a lower level to get to capital m or if you want to say moles when reading it

george: first theres the mathml and we need to make sure that the semantic information can be extracted from that notation
... and i think the role = chemistry is terrific

neil: i caution that that was a suggestion. it was not that everyone agrees. until the mathml community gorup comes up with a consensus
... in a more general situation i cant say you should be using that now

george: i think of this as an escape character taht tells you that now we are in chemistry
... so we need to get that into mathml
... then if there are places in the notation where the semantics are unknown or it could be a b or c, we need to figure out a way to improve the semantics of mathml there

neil: where you have some ambiguity you have to be able to do something to resolve that ambiguity
... which we have not come up with yet. i have been more pushing for progress on the core part. we have been having monthly meetings on general issues rather than core issues on what it can do in the browser
... there havent been meetings so far to take up semantic related issues

george: great that its already in the mathml group's scope. i understand why the core needs to be done first
... when we have one of these mathml general meetings, we need to have some of us join that call

neil: that would be great, especially if we start discussing things more in the weeds, down in the technical level

kerry: i would be happy to join that call for that purpose

dan: me as well

george: once we have that, then we will work on getting the screen readers and translators to go
... the folks from jaws said that they would love to have math and chemistry experts helping them because they need test material
... we can also do this with nvda

neil: we do hope to have test suites
... nothing above core yet

george: im sure we can get this to the microsoft narrator team. apple people are a little bit more harder to reach
... android talkback as well

dan: if i may, what exactly do you mean by test suites

neil: examples of math and examples of how they should be spoken

george: based on perfect input we should have predictable output

neil: often in math there are differnet ways to read things. i know in chemistry there are different ways as well. whether you say h2o versus water
... in mathplayer
... there may need to be different ways it read things in different settings

dan: can kerry and i start thinking about this ?

neil: absolutely. the mathml will be the same. its just a matter of different things being modify. if there is something wants to direct how they pronounce things (hydrocloric acid vs hcl), im not sure, we havent thought about that at all

kerry: some instructors would only want it to be pronounced one way or another. we'd want to be able to accommodate both

george: more examples: minus versus dash and h2o versus water
... people that use screen readers a lot understand that they are in a math context. they see a dash and know that it is a minus sign. but that adds a cognitive load

neil: any mathml to speech worth its salt would have turned that dash into a minus sign

bill: in effect we are discussing a pronounciation lexicon

george: we have the PLS in epub
... in the APA within the w3c there is a task force for pronunciation. this is very important in many contexts
... for example japanese has to be heavily annotated to get speech engines to pronounce things correctly.
... they have weekly meetings

steve: pronunciation is one aspect and the other more broadly are speech rules. that goes around to how mathplayer works to come up with how to actually speak
... specific terms
... as they fall
... mathplayer uses heuristics to determine whether a k is equilibrium or potassium or kelvin

bill: question for george: in those lexicons, is it a matter of pronunciation of a term, should it always be x or should it depend on the context?
... how will this get implemented in half a dozen ways. it would be nice that there would be a standard lexicon

neil: in the lexicon in epub, is it forcing a particular pronunciation? in math, there is the letter a (long a versus short a)
... but for something like recognizing K as potassium, is it really appropriate to use a lexicon for that?

bill: the simple example i remember has to do with acronym pronounciation. like a doc file but you say TSA, you dont say 'tsa'

george: proper nouns are frequently messed up. so i made a rule in my screen reader
... so the PLS lexicon in epub allows you to put in a whole llist of replacements that will speak it correctly. and that works great for many things like names. but in many cases
... it is context sensitive
... most screen readers and things like that figure out pretty well whether it is read or read (red)
... but not all the time. when you get into foreign languages, its a whole other thing. when the PLS lexicon works, it is case senstiive
... they are reocmmending in line mark up
... because of aversion to xml, they are looking at json
... the task force on pronunciation is trying to address these issues. they are having a meeting at TPAC
... i will be attending remotely
... this is super important in the testing arena where you have to say it correctly to the students taking these tests
... we need to attend the monthly meetings when chemistry is on the agenda
... we need to be prepared to put together sample materials that could be used in a test suite. we need to be prepared to look at what we think is the correct pronunciation of these things
... and there may be options
... and we need to be prepared to work with companies that are implementing these and help them out
... all of that is just in mathml and chemistry arena and there are a lot of other things that we need to be working on
... i think that is quite a bit of work moving forward. i would say that this is the most fundamental
... because mathml is so broadly implemented that that is good work we can do

bill: your comment about xml being out of fashion, which always bugs me, mathml is xml. so is there a way that ssml could work with mathml to enable that kind of thing

neil: there is way to do it but none of th translation would do it. there is a semantics element. you could have an ssml element that pronounces it. no one really uses it. none of hte implementations would pick up on that
... even though it is a standard, the speech engines are varied and some dont pick up ssml
... with the pronunciations it is often difficult to use. some might use IPA and some might have a propriatery way to pronounce it

kerry: the problem is getting screen reader companies to support SSML

neil: it is not clear to me that adding a lexicon is the best way to deal with pronunciation. to me it is about when a voice engine wont say sodium properly. you need to tell it how it is pronounce
... the language key also dictates the pronunciation so they get that right

george: many of the newer browsers have the read aloud function. for a person with a learning disability or dyslexia, we have to be thinking beyond screen readers and to the more general solution

bill: are any distinctions in unicode?

neil: not really. you can do things like mark it up without mathml for a number of cases in unicode and the mathml is undersatndable. but it doesnt dictate pronunciation.
... people shoudl already be using the proper unicode. its more about how it will pronounce multiple characters grouped together, like a fraction.
... the test suite examples you want are more edge cases
... something with potassium, kelvin, and so on would be a good example
... the mathml wont handle chemical structures. chemML or svg might work to encode that

george: volker isnt with us today but there is good work in that area that we want to build on but there is a ton of work in this area. part of it is
... the notation that is used. part of it is the presentation, probably in svg, and also the navigation of the structures and how they are voiced
... there is lots of work to do there as well
... we have the initial github repository that charles set up. he is on holiday. but i am feeling like we should move this over to the official chem-web-pub wiki and do all of work there
... is there anybody who is really on top of github and move that over. i dont think it would be too hard to move it over.
... just clone it and upload it to the new area
... when do we want to get to our next call?
... every fourth thursday we will meet
... getting into a regular cadence is great

This is scribe.perl Revision: 1.154  of Date: 2018/09/25 16:35:56  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/scrpt/script/
Present: George franco Bill_Kasdorf
Found ScribeNick: franco
Found Scribe: franco
Inferring ScribeNick: franco

WARNING: No "Topic:" lines found.


WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

WARNING: Could not parse date.  Unknown month name "06": 2019-06-27
Format should be like "Date: 31 Jan 2004"

WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

People with action items: item

WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report


WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)