WEBVTT Kind: captions Language: en 00:00:04.800 --> 00:00:10.240 Good afternoon and welcome. My name is  Anita Walz. I’m the open education,   00:00:10.240 --> 00:00:14.960 the assistant director for open education,  and scholarly communication librarian here   00:00:14.960 --> 00:00:20.560 at Virginia Tech. This event is sponsored by the  University Libraries open knowledge committee in   00:00:20.560 --> 00:00:26.240 consultation with colleagues from Virginia Tech's  office of accessible technology, undergraduate   00:00:26.240 --> 00:00:31.120 academic affairs, the College of Engineering,  the Department of Mathematics and Virginia Tech   00:00:31.120 --> 00:00:37.440 Publishing. Thank you so much for each of you  who were involved in planning this event for   00:00:37.440 --> 00:00:45.920 your involvement. This is the second of two events  in Virginia Tech's 2021 open education symposium.   00:00:47.200 --> 00:00:52.960 If you wish to find more information about  the other events, we have one more on Friday   00:00:52.960 --> 00:00:59.440 that is co-sponsored with SCHEV which is the  State Council on Higher Education for Virginia   00:00:59.440 --> 00:01:06.400 on Friday. And all of our recordings  will be available on the open education   00:01:06.400 --> 00:01:12.640 website for Virginia Tech. We welcome  seeing you or having you view the, the um,   00:01:14.560 --> 00:01:19.440 the information the, the recordings there after  they're posted those will be about a week.   00:01:21.360 --> 00:01:24.480 There are also many international virtual events   00:01:24.480 --> 00:01:28.240 and there's more information  available at openeducationweek.org. 00:01:30.000 --> 00:01:34.880 The idea for this presentation came out  of multiple efforts to make STEM content,   00:01:35.760 --> 00:01:40.720 specifically math rich open educational  resources accessible to people with low vision.   00:01:41.280 --> 00:01:45.520 Over the last five years I have, with  colleagues in Virginia Tech publishing,   00:01:45.520 --> 00:01:51.200 experimented with several different methods for  making math intensive educational resources,   00:01:51.200 --> 00:01:56.960 open educational resources accessible. These  have involved receiving manuscripts in print,   00:01:56.960 --> 00:02:02.080 LaTex and Microsoft Word. We have  worked with PDF tagging, LaTex,   00:02:02.080 --> 00:02:08.000 and MathML in press books as well as  some other methods. We participated   00:02:08.000 --> 00:02:14.000 and initiated conversations with many of you on  this call. In designing this program, we have two   00:02:14.000 --> 00:02:21.200 main two overarching goals in mind. We hope  that digital math, STEM, and, and open education   00:02:21.200 --> 00:02:26.080 communities will catch a vision for what is  possible regarding STEM and accessibility,   00:02:26.880 --> 00:02:32.160 and we hope you'll be motivated to learn more.  We've created an annotated list of resources   00:02:32.160 --> 00:02:38.560 that we hope you will add to this is available  at bitly/ accessible_STEM. That link should   00:02:38.560 --> 00:02:43.520 also be in the chat. We hope that you'll gain  greater insight into the importance of making   00:02:43.520 --> 00:02:48.880 and sharing content which is not only openly  licensed, but is also accessible with that. 00:02:48.880 --> 00:02:54.720 I'd like to introduce you to our speaker. Dr  Volker Sorge is professor in document analysis   00:02:54.720 --> 00:02:59.120 and accessibility at the School of Computer  Science at the University of Birmingham in   00:02:59.120 --> 00:03:05.520 the UK. Dr Sorge leads the scientific document  analysis group whose research primarily focuses   00:03:05.520 --> 00:03:10.960 on mathematical document analysis, diagram  recognition, and handwriting recognition.   00:03:11.760 --> 00:03:14.560 He's worked on STEM accessibility for many years.   00:03:15.280 --> 00:03:21.040 As a visiting scientist at Google. He integrated  math support into the ChromeBox screen reader.   00:03:21.920 --> 00:03:26.320 As a member of the MathJax consortium he has  worked on formula accessibility on the web.   00:03:26.880 --> 00:03:31.840 Most recently, in collaboration with the Pretext  Group, the National Federation for the Blind,   00:03:31.840 --> 00:03:37.280 and the American Institute of Mathematics, he  contributed to the automatic transcription of   00:03:37.280 --> 00:03:43.680 LaTex textbooks into tactile braille books. In  his startup company, Progressive Accessibility   00:03:43.680 --> 00:03:48.880 Solutions, Volker focuses on exploiting  pattern recognition and image analysis   00:03:48.880 --> 00:03:54.240 technology for automatically making STEM diagrams  accessible for use in teaching and science   00:03:55.120 --> 00:03:59.600 you study mathematics computer science  and philosophy at the University of  00:03:59.600 --> 00:04:03.920 Würzburg, University of Texas at  Austin, and the Universität des   00:04:04.560 --> 00:04:10.640 Saarlandes. He holds research Masters in  Mathematics, and a PhD in computer science   00:04:10.640 --> 00:04:17.520 in computational logic. Please give a warm,  socially distanced welcome to Dr Sorge. 00:04:17.520 --> 00:04:20.720 Hello; thank you. Thank you very  much for the kind introduction.   00:04:21.360 --> 00:04:26.800 Um yes, thank you again for the kind introduction.  So, my name is Volker Sorge. As, uh Anita said,   00:04:26.800 --> 00:04:33.280 I mm have a professorship at the University of  Birmingham, um, where I’m nominally now, um,   00:04:33.280 --> 00:04:37.360 lead the scientific document analysis group,  because most of my time I actually spend,   00:04:37.360 --> 00:04:42.880 um elsewhere not just because we're still in full  lockdown and therefore the university is closed,   00:04:43.760 --> 00:04:47.600 but also because I’m primarily now working for my  own little company, which is called Progressive   00:04:47.600 --> 00:04:53.120 Accessibility Solutions, as well as for the  MathCheck consortium. And as you can imagine,   00:04:53.120 --> 00:04:58.720 I will talk quite a bit today about MathCheck  uh in the course of this presentation.   00:05:00.160 --> 00:05:06.160 And um I’m very glad to have been invited for this  particular event because I think it's an important   00:05:06.160 --> 00:05:12.000 subject, accessibility, and it's particularly  important right now at this particular point in   00:05:12.000 --> 00:05:17.280 time, and I shall motivate a bit throughout  the beginning of the talk why first of all   00:05:17.280 --> 00:05:23.360 accessibility is important, why it is important at  the current moment and also particularly why it's   00:05:23.360 --> 00:05:32.560 important in education and particularly in higher  education for STEM subjects. And um I will, so,   00:05:32.560 --> 00:05:38.640 towards the second part of my talk, I will then  talk a bit about, a bit about the science behind   00:05:38.640 --> 00:05:43.440 some of these techniques that we're working  on as well as some of the technology that's   00:05:43.440 --> 00:05:48.400 already available and hopefully that will become  available in the future to, to help everybody uh   00:05:48.400 --> 00:05:54.080 with making their content accessible and therefore  also helping people to actually access it. 00:05:55.440 --> 00:05:59.280 So, first question is of course what  is accessibility? Can we somehow define   00:05:59.280 --> 00:06:05.920 accessibility? Um well first of all what  is it about, in particular in our context.   00:06:05.920 --> 00:06:09.920 Most people when you tell them, well,  accessibility they think maybe about uh   00:06:09.920 --> 00:06:16.960 ramps being built to buildings or alternatively  have large cubicles and in toilets um,   00:06:16.960 --> 00:06:22.640 but when we talk about accessibility, we  mainly talk about accessibility, um, which   00:06:23.280 --> 00:06:30.160 is about overcoming barriers that we build by  building more and more modern electronic gadgets   00:06:30.720 --> 00:06:36.400 and there's quite a number. So, everybody in many  ways has some sort of a disability at some point,   00:06:36.400 --> 00:06:41.280 so there's obviously people that are  born or, or have had a disability   00:06:41.280 --> 00:06:46.720 due to a late accident in life like visual  disabilities. People are blind, low vision,   00:06:46.720 --> 00:06:52.080 but there's other things to consider also. You  know people that have are deaf, that have aural   00:06:52.800 --> 00:06:58.320 distractions, people that cannot use certain  gadgets like because they forget this, can't   00:06:58.320 --> 00:07:05.760 use a mouse um, they have to they have to use um  they have other limited modified motor skills they   00:07:05.760 --> 00:07:12.880 can only for instance access a computer with  their mouth etc., and various other cognitive   00:07:12.880 --> 00:07:17.680 disabilities which particular interesting I  always find in a university setting which have   00:07:17.680 --> 00:07:23.400 to do with things like dyslexia, discalcula,  etc. for which we're trying to provide, ah,   00:07:24.160 --> 00:07:31.120 something that overcomes these barriers. And, uh,  just I always like to say, well, everybody's not   00:07:31.120 --> 00:07:38.400 fully able at some point, everybody's is not fully  able at some point in our life so um we all have   00:07:38.400 --> 00:07:46.080 some temporal situational disabilities at some  point. So, for instance if you work outside and   00:07:46.080 --> 00:07:50.160 there's a lot of sunshine all of a sudden you  have glare on your screen then you, you know   00:07:50.160 --> 00:07:55.280 you might want to, want to change the setup there,  or aren't we all glad, for instance, that when we   00:07:55.280 --> 00:08:02.720 drive our car, that the GPS automatically changes  to night colors, uh when the sun goes down. And,   00:08:02.720 --> 00:08:08.000 and all these are in many ways disabilities  and disabilities we have to take account for. 00:08:10.400 --> 00:08:16.640 And in particular important this is in, in  the field of inclusive education because   00:08:18.160 --> 00:08:22.320 equal access to education  is an important human right,   00:08:23.120 --> 00:08:27.680 and everybody should have access to  education regardless of their special needs.   00:08:29.040 --> 00:08:35.120 And consequently, it is important that in  the teaching setting accessibility is, is,   00:08:36.320 --> 00:08:43.680 it is guaranteed. And again, I’m not talking  about wheelchair accessible lecture theaters;   00:08:43.680 --> 00:08:49.760 I’m talking about the actual teaching material  that has to be accessible. It's no good if I can   00:08:49.760 --> 00:08:55.360 get into the lecture theater and find a seat,  if I then cannot participate in the lecture   00:08:55.360 --> 00:09:00.640 because I simply cannot take, I cannot use  the same material as, as my peers around me.   00:09:02.320 --> 00:09:10.240 As a consequence, there has been, you know,  not, not just now, but for ,for centuries as   00:09:10.240 --> 00:09:15.920 a matter of fact, but particularly over the  last 50-60 years, there's been a lot of work   00:09:15.920 --> 00:09:21.840 on providing things like alternative formats and  specialist help for people in education setting to   00:09:22.400 --> 00:09:28.880 ensure inclusive education, but often these  these, um, efforts are made mainly in, in um,   00:09:29.520 --> 00:09:35.520 in primary secondary school so k to 12 you  would say in the US. Um, and then for somehow,   00:09:35.520 --> 00:09:41.120 somehow once the university stage is  there, you know, it becomes less uh   00:09:42.000 --> 00:09:49.840 so the, the support becomes less widely  available for, for students with special needs.   00:09:50.480 --> 00:09:55.760 So, ahem, we're trying to change this,  obviously. And how are we trying to change it. 00:09:55.760 --> 00:10:00.480 Well let's first have a look at what is  the historical solution or what are the   00:10:00.480 --> 00:10:05.680 historical solutions to these problems and  then let's see why these don't work anymore   00:10:05.680 --> 00:10:09.360 and how we can change them  in for a modern setting.   00:10:10.800 --> 00:10:16.560 So historically you would make, or you would  give students with special needs alternative   00:10:16.560 --> 00:10:21.280 material. When I say alternative material that  means instead of saying, say, giving them a   00:10:21.280 --> 00:10:27.440 book which is printed you might give a blind  student a book that is embossed in braille.   00:10:28.640 --> 00:10:34.080 Likewise, you can give them an audio book, you can  give them a textbook that has been read out by an   00:10:34.080 --> 00:10:39.120 expert and they can listen to it. They can rewind  things and they can thereby follow their book.   00:10:43.520 --> 00:10:47.680 Engineering students for instance, but also  students in mathematics, they would then get   00:10:47.680 --> 00:10:54.800 particular embossed diagrams in  order to follow graphical content.   00:10:54.800 --> 00:11:00.400 So tactile prints can be prepared as well;  paper can be used so things are automatically   00:11:00.400 --> 00:11:05.040 being raised if you draw on them, and then  they can experience those if they are blind.   00:11:07.040 --> 00:11:12.480 Likewise, complex content then can often  be spoken out by expert humans and can   00:11:12.480 --> 00:11:18.320 explaneed to them in more details, if you have  a video, you can subtitle it for, for students   00:11:18.320 --> 00:11:27.040 with aural or with hearing impairments and so  on and so forth. And likewise, in, in exams,   00:11:27.040 --> 00:11:32.160 people would get specialist support, they would  get their nuances if they can't write and so on.   00:11:33.440 --> 00:11:37.440 Um, these days this is a lot harder to  provide, because there's a couple of   00:11:38.240 --> 00:11:43.920 reasons for that. The first reason is, of course,  that now a lot of content is in electronic form,   00:11:43.920 --> 00:11:48.640 and not only is it an electronic form  but it's also being prepared on the fly. 00:11:48.640 --> 00:11:54.000 So, I’m aware that in the US you still use, or  people still use, textbooks but for instance in my   00:11:54.000 --> 00:12:00.320 own lectures, for many years I've told students,  well, I provide handouts and the handouts are good   00:12:00.320 --> 00:12:05.440 enough. You can have secondary reading but there's  no need for a specialist textbook anymore. Now the   00:12:05.440 --> 00:12:10.080 that is in many ways great, because students don't  have to buy a textbook anymore. On the other hand,   00:12:10.080 --> 00:12:15.120 what is the effect is, that these are my handouts,  my lectures. I can change them the night before   00:12:15.120 --> 00:12:22.640 the lecture and I often do. As a matter of, as,  as a matter of fact this is not just me. I know   00:12:22.640 --> 00:12:28.640 that happens a lot with other students and we are  always told by our accessibility support services,   00:12:28.640 --> 00:12:34.320 well, provide certain students at least 48 hours  in advance with the, with the teaching material,   00:12:34.320 --> 00:12:40.000 but often that doesn't happen. Similarly, if  you need to, if you have a blind student in   00:12:40.000 --> 00:12:46.000 your class who really needs that material in an  alternative format how you can you translate it   00:12:46.640 --> 00:12:54.080 by the time of the lecture if you've just changed  the lecture notes the night before? So, all these   00:12:54.080 --> 00:12:59.760 provisions that used to work historically  don't really work anymore in the light of   00:13:00.480 --> 00:13:06.640 ephemeral material like teaching material  on the web and it's getting even worse. 00:13:06.640 --> 00:13:12.080 So, what about COVID 19? Now all of a sudden, it's  not just the teaching material that we put into   00:13:12.080 --> 00:13:18.000 learning management systems. Now everything has  to move online, right? It's not just the handouts,   00:13:18.000 --> 00:13:22.560 but it's also the teaching, the entire  learning experience. Exams are online.   00:13:22.560 --> 00:13:28.000 Even the social interaction, so that means  that even more barriers that are being built.   00:13:28.000 --> 00:13:33.520 Because, while maybe previously a student  who had some problems reading something or   00:13:33.520 --> 00:13:39.600 couldn't hear something properly could ask a  peer whether they could translate them for it can   00:13:39.600 --> 00:13:43.120 translate it for them or could  help them with a certain material,   00:13:43.120 --> 00:13:48.000 now that's no longer possible because everything  is social distant. They all sit in their own   00:13:48.000 --> 00:13:55.440 little dorm room and that's it. And likewise, no  experts can read them formulas and there is no way   00:13:55.440 --> 00:14:00.960 of an amanuensis is coming to help them writing  their exams. And for us as well, all of a sudden,   00:14:00.960 --> 00:14:06.800 we have to subtitle lecture videos now. That's,  you know, if it works, it's great. Unfortunately,   00:14:06.800 --> 00:14:16.000 many subjects, in particular STEM subjects,  automatic subtitling is not really an option yet. 00:14:16.000 --> 00:14:22.400 So, we're in a pandemic. Does that mean we should  forget about accessibility, or have it as an   00:14:22.400 --> 00:14:29.120 afterthought? And, and obviously my obvious answer  is no, we shouldn't. And why shouldn't we? Well,   00:14:29.120 --> 00:14:34.080 because it's important for the people that  need it, but it's also important for us   00:14:34.080 --> 00:14:39.040 that do it in a way. Because there's a couple of  questions I want to debunk here. So, for instance   00:14:40.400 --> 00:14:46.160 is there always the usual replies of why don't  we have more pressing matters now? Don't we   00:14:46.160 --> 00:14:51.920 have to move things fast onto the web? Doesn't it  focus down trying to make everything accessible?   00:14:52.800 --> 00:14:58.000 And my answer here is, of course, no it  doesn't. Actually, it's the opposite,   00:14:58.000 --> 00:15:04.720 because if you now do a rush job, all that  means is that you put bad material online   00:15:04.720 --> 00:15:10.400 and eventually it comes back and haunt, to haunt  you and you still have to put in the effort of   00:15:10.400 --> 00:15:17.360 making it accessible and it, it doesn't really  go all that much faster, because if you know   00:15:17.360 --> 00:15:22.160 what you're doing and if you just put in this  little epsilon, this little extra bit of work   00:15:23.280 --> 00:15:30.720 right now, then you get not only accessible  material, but also future-proof material,   00:15:30.720 --> 00:15:33.760 and material that you do not have  to constantly go back and revise. 00:15:36.000 --> 00:15:42.080 And next question is, of course, always, well  isn't the target audience too small? Um well,   00:15:43.040 --> 00:15:47.760 no, because the target audience ultimately is  everybody. If good material is online, it's   00:15:47.760 --> 00:15:52.480 good for all students; it's good for everybody's  experience. It's also good for the lecturer who   00:15:52.480 --> 00:15:57.200 doesn't have to constantly revise their lecture  notes. Moreover, if you think about all the   00:15:57.920 --> 00:16:02.320 tools and, and over the years that  have been originally being built,   00:16:02.320 --> 00:16:09.440 been built just to help and to support people  with disabilities and we all use now, um. 00:16:10.240 --> 00:16:14.640 Just as an example, you know, think  about something like speech output   00:16:15.920 --> 00:16:20.880 that used to be particularly for blind, for  blind users of computers. Nowadays we ask   00:16:20.880 --> 00:16:26.080 Alexa and we're happy that you reply to us  with the weather forecast for tomorrow. Um,   00:16:27.520 --> 00:16:31.920 other things, uh, here's a little example  which I believe everybody can see that we   00:16:31.920 --> 00:16:38.960 can all benefit from it. So, here's a bit, bit  I've stolen from the Wikipedia page on functions,   00:16:38.960 --> 00:16:46.880 but it's just a, um, a regular graph and it has,  it has a caption which says the red curve is the   00:16:46.880 --> 00:16:51.920 graph of a function because any vertical line  has exactly one crossing point with the curve.   00:16:51.920 --> 00:16:55.600 Now here's a couple of things that we can  observe. First of all, if we look at that   00:16:56.480 --> 00:17:02.320 this image, it's a bit grainy and, you know,  maybe I want to have it slightly larger. Let's   00:17:02.320 --> 00:17:07.760 make it slightly larger. What we can see here is  it gets worse, the graininess, if you, if you,   00:17:07.760 --> 00:17:12.480 you know. Already I have problems discerning  that this here is a seven, that this is an eight,   00:17:12.480 --> 00:17:17.520 right? And the reason for this is because the  format of the image is just poorly chosen. This   00:17:17.520 --> 00:17:24.080 is a bitmap image. Bitmap images do not scale. So,  what do we do about that? Well, let's have a look   00:17:24.080 --> 00:17:30.560 at a different version of the same image. Well,  here that's an SVG, so scalable vector graphic.   00:17:30.560 --> 00:17:38.000 The advantage here is that it scales and that  is accessible to some degree and if I, if I zoom   00:17:38.000 --> 00:17:43.840 that in now because I want to see a bit more,  now I can see actually oh this 0.8 here becomes   00:17:43.840 --> 00:17:50.400 it comes out very nicely there's no graininess.  Everything scales nice if you look at that graph. 00:17:51.360 --> 00:18:01.280 Now one question you might have is where's the  x-axis? Well, somebody decided that to make the x,   00:18:01.280 --> 00:18:06.400 oh no, sorry, the x-axis actually, the y-axis;  where's the y-axis? Somebody decided to make   00:18:06.400 --> 00:18:12.160 the y-axis, um, yellow for some reason. On  white background, that's a bit nasty isn't it,   00:18:13.280 --> 00:18:20.160 um, so it would be nice if this were in a  different color. Um, and there's something else   00:18:20.160 --> 00:18:26.240 terribly wrong with the whole, with the whole page  here. And I’ll give you a hint if you don't know   00:18:26.800 --> 00:18:32.080 what it is yet. If you read the caption  again, it says, well, the red curve is   00:18:32.080 --> 00:18:38.080 the graph of the function and I have problems  seeing the, the yellow line. So, what I do,   00:18:38.080 --> 00:18:44.800 well, I use a little extension of my browser which  simply gives me a high contra, contrast value,   00:18:46.320 --> 00:18:51.680 uh, high contrast color values. And let's,  let's just shift 11, which I press now,   00:18:51.680 --> 00:18:57.920 and what do I get? Oh well, all of a  sudden, I can see the y-axis now it is blue   00:18:57.920 --> 00:19:03.760 and, unfortunately, now I can read the caption  again. The red curve is the graph of the function.   00:19:04.880 --> 00:19:08.800 Well, what is the red curve? The red curve  is gone. The red curve is now blue. That's a   00:19:08.800 --> 00:19:12.880 bit of a nuisance, and the other, the only  other thing I can see is now that I have   00:19:12.880 --> 00:19:18.800 a yellow vertical line here maybe that refers  to the x-axis, but probably not. But yeah, um,   00:19:20.080 --> 00:19:25.200 so how do I solve the dilemma? Well let's go to the next, next page and have a look   00:19:25.200 --> 00:19:29.760 how that looks like. Huh, well,  let me let me switch this off   00:19:30.800 --> 00:19:34.640 now. I've written the red curve,  at least I put red around it,   00:19:35.280 --> 00:19:43.520 and then, and that refers to the same color as the  curve. And if I now switch on my, if my, my high,   00:19:44.400 --> 00:19:50.480 high contrast colors, at least I can get some  sort of a reference between what the red curve is   00:19:51.040 --> 00:19:56.720 and what the actual color here is. Obviously,  that's still not ideal. Ideal is a scenario   00:19:56.720 --> 00:20:02.800 like this, where this is also stolen from the same  page, uh, with a slightly extended caption, where   00:20:02.800 --> 00:20:09.360 I see two trigonometric function and I can discern  the sign not only by its color but also because   00:20:09.360 --> 00:20:15.520 it's a solid curve, whereas the cosine is a blue  dashed curve. And if I change the colors now,   00:20:15.520 --> 00:20:20.720 then obviously blue and red don't really work  anymore, but at least solid and dashed still do. 00:20:24.160 --> 00:20:31.760 So, this is just an example, um, why these things  are quite useful not only for people with visual   00:20:31.760 --> 00:20:35.680 impairments, but pretty much for everybody. When  you have, when you have problems with contrast   00:20:35.680 --> 00:20:40.960 and, and when you maybe are different or  prefer a different a layout of colors.   00:20:42.880 --> 00:20:49.520 Um now these are in many ways, what I've just  shown you are certain properties that make content   00:20:49.520 --> 00:20:55.520 accessible, if they're there or if they're not,  right? If you, if you, if you refer to something   00:20:55.520 --> 00:21:01.360 by color only and that color changes, it's clearly  no longer accessible for the person who has,   00:21:01.360 --> 00:21:08.800 who has to use a different color palette for  because of their visual impairment. Um, so there's   00:21:10.480 --> 00:21:15.600 what's called the web accessibility or content  accessibility guidelines: the Woocock Guidelines,   00:21:15.600 --> 00:21:21.360 which are generally accepted standard, not  just for the web, but for, for many electronic   00:21:21.360 --> 00:21:28.720 material, um, how things should be and how they're  being made, should be made accessible. Many of   00:21:28.720 --> 00:21:35.520 these accessibility guidelines inform countries  in, in, in most countries and formed, not   00:21:35.520 --> 00:21:42.080 countries sorry, inform the law on accessibility  in most countries. Some of them are optional or   00:21:42.080 --> 00:21:49.600 ideal, um, but I, I mean it's pages and pages of  these guidelines, roughly 60 success criteria,   00:21:50.400 --> 00:21:57.200 and the main idea is what they call the Poor  Principle. It's about content should be for   00:21:57.200 --> 00:22:03.120 all users perceivable, it should be operable,  it should be understandable, and it should be   00:22:03.120 --> 00:22:08.880 robust. So perceivable means that if I can't see  the image, at least I need to have some sort of an   00:22:08.880 --> 00:22:15.440 alternative description. If I can't listen to the  video, at least I need subtitles. Operable means   00:22:15.440 --> 00:22:20.480 if I can't use a mouse, for instance, I still need  to be able to get to certain elements by keyboard,   00:22:21.920 --> 00:22:27.440 and if I understandable means, of course,  it's about the language of the document,   00:22:27.440 --> 00:22:32.400 but often also about something like  referring to things like color,   00:22:32.400 --> 00:22:36.720 because all of a sudden it does no longer become  understandable what curve you're talking about,   00:22:37.760 --> 00:22:44.160 and then robustness of, of course is that things  should not just be optimized for this particular   00:22:44.160 --> 00:22:47.520 browser on this particular operating  system, but it should work everywhere.   00:22:52.160 --> 00:22:57.840 And, uh, just as a and a little aside, because  I was talking about the operability, so   00:22:58.400 --> 00:23:05.840 I occasionally gave, gave bootcamps for  students so they get starting software engineers   00:23:05.840 --> 00:23:10.320 and so they get an idea what accessibility is all  about. So, the first thing I normally do is I,   00:23:10.320 --> 00:23:15.120 I plug, I, I take away their mice  and sit them in front of a computer.   00:23:15.120 --> 00:23:18.000 They're always quite shocked; how  they can even log in without a mouse?   00:23:18.720 --> 00:23:22.880 And the second thing then is you switch off  their screen and switch on a screen reader   00:23:22.880 --> 00:23:28.000 so they can really experience how it is if  you can't see what you put in front of them.   00:23:30.880 --> 00:23:34.080 So, this is general accessibility. 00:23:34.080 --> 00:23:38.480 What does that mean for STEM? what does  that mean for science, technology and math?   00:23:41.760 --> 00:23:50.160 Well, the problem is, even if all these  criteria are really, um, fulfilled or satisfied,   00:23:50.160 --> 00:23:54.400 that, that still doesn't make the actual STEM  content accessible, and that has a couple   00:23:54.400 --> 00:23:59.440 of reasons because, first of all, of course  STEM is quite different from regular document   00:23:59.440 --> 00:24:06.000 content in documents. And it's also hard what  math and other science documents bring together   00:24:06.640 --> 00:24:11.600 are most of the hardest accessibility problems.  So, for instance, you have specialized vernacular,   00:24:11.600 --> 00:24:16.080 very specialized things you can't just  look up in dictionaries. You have scripts,   00:24:16.080 --> 00:24:22.000 um, that might be in the mix. So, you have  three characters in between Latin letters,   00:24:22.000 --> 00:24:24.960 sometimes even in the same  word. Think about chemical, uh,   00:24:25.600 --> 00:24:33.600 components or chemical, um, molecules described in  IUPAC form. They have alphas and parentheses and,   00:24:33.600 --> 00:24:40.480 and, and type punctuation inside words. You have  tables, formulas, diagrams. Modern days you might   00:24:40.480 --> 00:24:46.800 have animation simulations and whatnot. Everything  all together in single pieces of content.   00:24:48.080 --> 00:24:54.240 And another problem we have in with, with STEM  is that effectively, STEM are infinite languages   00:24:54.880 --> 00:24:59.040 and that's the whole point about science,  right? We constantly enrich the language,   00:24:59.040 --> 00:25:06.080 we constantly invent new concepts, we constantly  define new stuff. We give them new names so   00:25:06.080 --> 00:25:10.000 it's not like in English where, you know, an  Oxford Dictionary has to be revised every few   00:25:10.000 --> 00:25:18.000 years adding a few words, taking a few out. In,  in, in scientific languages this is not enough,   00:25:18.000 --> 00:25:22.800 because it grows every day, meaning that you  can't really declaratively prescribe things,   00:25:22.800 --> 00:25:28.720 and all these standard accessibility techniques  that work for standard content really fall flat.   00:25:30.640 --> 00:25:36.400 And even if solutions exist, most solutions  are really just good for k-12, um,   00:25:38.240 --> 00:25:42.160 because that's, let's face it, this, this  is where the commercial market is. But it   00:25:42.160 --> 00:25:46.960 doesn't have to be that way, and that is  one of the things I often get asked. And,   00:25:46.960 --> 00:25:55.840 and I would like to debunk a couple of standard  and common myths about STEM accessibility. 00:25:58.880 --> 00:26:03.120 So, one is for instance that people think  that PDF, once I put a PDF document on the   00:26:03.120 --> 00:26:07.360 web it's accessible. People should be able to  deal with this. It's definitely not the case.   00:26:07.920 --> 00:26:12.640 It's very rarely accessible for regular text.  It is definitely not accessible from when you   00:26:12.640 --> 00:26:19.200 have mathematical content in there. Just try to,  to copy out a mathematical formula out of the,   00:26:19.200 --> 00:26:26.560 out of a PDF document and see what garbage you  get on your, on your console, on your text editor. 00:26:28.960 --> 00:26:34.000 Um, another problem is that, um, people often  about being told, well, you have to entirely   00:26:35.200 --> 00:26:40.240 change how you work. We want everything from you  in Word and MathML, otherwise we can't make it   00:26:40.240 --> 00:26:46.560 accessible. That's also not true. By all means, do  not change your authoring habits. Do not move away   00:26:46.560 --> 00:26:52.960 from things like LaTex etc. As a matter of fact,  LaTex is so much more preferred to things to,   00:26:52.960 --> 00:26:58.240 to standards like MathML, because it's  much richer and one can do a lot more in   00:26:58.240 --> 00:27:03.840 terms of understanding the mathematics that's  being expressed. What is important, though,   00:27:03.840 --> 00:27:11.680 is that people understand certain requirements of  accessibility and, therefore, get rid of certain   00:27:11.680 --> 00:27:16.640 bad habits when they author. And that's something  that people can learn. And once they learn it,   00:27:17.440 --> 00:27:22.080 they become much more proficient at  providing good content and good material. 00:27:25.280 --> 00:27:31.200 And another one is always that people have,  well, my, my content the content I produce   00:27:31.200 --> 00:27:35.680 is far more too, far too complex to make it  accessible. That's also not true. Generally,   00:27:35.680 --> 00:27:41.280 it just means that the way it's authored is  too, too complex, not the content itself. 00:27:43.440 --> 00:27:50.080 And I’ll tell you how to make some of that content  accessible. And we'll start effectively with just   00:27:50.080 --> 00:27:58.640 regular, with just regular documents, mathematics  documents. And in order to understand a bit how   00:27:58.640 --> 00:28:05.840 the accessibility really works, it's important  to understand a bit how electronic content is   00:28:05.840 --> 00:28:12.080 different from traditional content, how the web  in particular, but also other electronic content,   00:28:12.080 --> 00:28:22.560 um, say on, on eBook readers etc. But even to  some extent um in in in formats like PDF or DVI,   00:28:23.280 --> 00:28:31.360 the content is different. Because what happens  is that the content is being provided in a way   00:28:31.360 --> 00:28:38.480 that it's being rendered somewhere else, right,  so one previously in, in previous days, you would   00:28:38.480 --> 00:28:43.600 have a camera-ready copy, you would put it onto  a printer. And then somebody would get a book,   00:28:43.600 --> 00:28:47.440 meaning that the content is fixed, the  rendering is fixed once it's delivered. 00:28:49.200 --> 00:28:53.360 Nowadays it's different. We deliver the content  first and then it's being rendered in the browser.   00:28:53.360 --> 00:28:58.480 It's being rendered in, in your PDF reader,  reader. It's being rendered somewhere else.   00:29:00.480 --> 00:29:06.880 A lot of the what the content electronic  content generally is a vertical content as   00:29:06.880 --> 00:29:13.680 opposed to traditional content which is, uh,  horizontal, right? You turn the page nowadays,   00:29:13.680 --> 00:29:19.840 we scroll down and, just to give you an idea how  that really works with the mathematical content,   00:29:22.320 --> 00:29:27.840 this here is a math paper from the art side  on a disproof of conjecture by Rademacher.   00:29:28.800 --> 00:29:35.360 And although this looks like PDF and it originally  was, it came from PDF this is really not PDF;   00:29:35.360 --> 00:29:42.080 this is a web document. However, what happened is  that we've translated it with a PDF to HTML and   00:29:42.080 --> 00:29:50.960 all these things are little spans in here and, and  therefore the, the entire pagination, the entire   00:29:50.960 --> 00:29:57.200 layout, everything is preserved. It doesn't  really look like a web document, as I said,   00:29:57.200 --> 00:30:05.920 but it is, all right. It's just done as if it had  pages. Right, but that's not really what, what new   00:30:05.920 --> 00:30:12.000 modern formats are really all about, yes? Let's  get rid of this. Let's look at the same documents   00:30:12.000 --> 00:30:17.280 translated slightly differently. All right, so  here it is, here's a live document, just on a web   00:30:17.280 --> 00:30:22.880 page. It's still the same, it's still the same  article. It's translated slightly differently,   00:30:22.880 --> 00:30:29.520 and now you can see something which is  common with, uh, with electronic documents,   00:30:30.080 --> 00:30:33.840 good electronic documents, that you can  actually reflow the whole thing. So,   00:30:33.840 --> 00:30:40.640 I zoom in a bit. Oops now I zoomed out. And  you could, you can start seeing the reflow   00:30:42.240 --> 00:30:48.640 of, of, of the text, of a certain, the text  of the next page and so and, and therefore,   00:30:48.640 --> 00:30:55.680 you can also see what I meant with the, the  content or, or the medium is a vertical medium.   00:30:55.680 --> 00:31:00.480 All we ever want to, really do is scroll up and  down. Nobody wants to scroll or pan sideways. 00:31:01.440 --> 00:31:07.840 All right, let's, let's go a bit further.  And also, this year now, you know,   00:31:07.840 --> 00:31:13.920 you've got live links, um, and you've got a  live formula and “oops, something's happening”   00:31:14.480 --> 00:31:20.960 um, it's actually interactive. We can have a look  at the original, uh, TeX command which produced   00:31:20.960 --> 00:31:26.000 that formula. I can copy it out; I can take it;  I can immediately plagiarize it, put it into my   00:31:26.000 --> 00:31:30.800 own paper do whatever I like with it. And this is  pretty much what you want from a web document or   00:31:30.800 --> 00:31:38.480 from, from a, from an electronic document, from  a flexible document. And how do we get that.   00:31:43.200 --> 00:31:49.920 Well, next, the important thing is to  understand how to transform mathematics   00:31:49.920 --> 00:31:55.440 so it's nice and that it's a good top becomes a  good document that's, uh, that's really useful. 00:31:55.440 --> 00:32:00.480 So, I’m going through the, the GBU, so the good,  the bad, and the ugly of the math content that's   00:32:00.480 --> 00:32:05.280 out there. And all of this can be produced, and  I’m telling you which one you should produce and   00:32:05.280 --> 00:32:09.840 which you definitely should not produce. So, let's  start with those I don't want you to produce ever.   00:32:12.000 --> 00:32:18.320 So, this is a web page which is, uh, MathWorld.  Let's zoom out a bit. So, this is MathWorld and   00:32:18.320 --> 00:32:23.840 it's got some really old content on here. This  is quadratic formula, of course, and what you can   00:32:23.840 --> 00:32:29.520 see here this is a picture, this is an image, a  bitmap image similar to that graph we had earlier,   00:32:29.520 --> 00:32:35.840 and it's really grainy. And once you, once  you need a higher assume factor or, or go in a   00:32:35.840 --> 00:32:41.840 different form factor, you can see this is really,  really ugly and hardly to read, hard to read. 00:32:42.560 --> 00:32:46.960 Now what you can do, somewhat something  else also somewhat old-fashioned, is   00:32:46.960 --> 00:32:53.600 you can style it in some way, um, that  is being used in just natural HTML.   00:32:54.320 --> 00:33:00.400 And here's another example of the quadratic  formula. And this here now is in in HTML,   00:33:01.760 --> 00:33:08.800 and as you can see, it's quite ugly right? It  certainly doesn't have any LaTex quality to it   00:33:08.800 --> 00:33:15.600 whatsoever. The radix doesn't even fit the  overbar here, but at least it's not an image;   00:33:15.600 --> 00:33:21.920 at least you can grab things and you can possibly  copy things out, um, but you can also imagine that   00:33:21.920 --> 00:33:25.520 this doesn't really work as soon as the  formulas become larger and more nested. 00:33:28.480 --> 00:33:33.760 All right, slightly more modern, then, is if,  if you render it like, like they do on Wikipedia   00:33:33.760 --> 00:33:40.400 these days. Here's the quadratic formula in  Wikipedia. What this here is, this here is an SVG,   00:33:41.120 --> 00:33:46.240 but again it is an SVG image, so I  can crop it and I can move it around.   00:33:47.280 --> 00:33:52.880 There's nothing in the else to do and, and  but at least it zooms nicely so it doesn't,   00:33:52.880 --> 00:33:59.280 you don't get any checkies you, don't  get any blur. And let's, let's zoom back. 00:34:01.440 --> 00:34:06.480 And then finally, of course, what you, what we  ideally want that content to be; you want the   00:34:06.480 --> 00:34:11.920 content to be fully rendered. Sorry, my slide  jumped away. So, here's fully rendered content   00:34:14.720 --> 00:34:20.400 and that's our on our own MathCheck page.  Again, the quadratic formula here, it is right   00:34:20.400 --> 00:34:26.720 fully rendered looks a bit like the one on the  on, on, on Wikipedia, but it isn't because I like,   00:34:26.720 --> 00:34:32.400 if I go there, if I focus on it, I press  return, things start to happen. All of a sudden,   00:34:32.400 --> 00:34:37.520 I get a magnified version of the formula. I get  some subtitles which actually give me text which   00:34:37.520 --> 00:34:42.960 I can read through and I can, I, you know, things  are being highlighted and, as a matter of fact,   00:34:42.960 --> 00:34:48.000 I can do more things. I can use my cursor  keys. I can start navigating this whole thing   00:34:48.000 --> 00:34:53.600 and I can navigate bit by bit, and things are  changing. The subtitles are changing the facts,   00:34:53.600 --> 00:34:58.560 the magnification is changing, and this  is really now an accessible formula.   00:34:58.560 --> 00:35:05.360 This is accessible for people not just for you and  me who can see this thing, but also for people who   00:35:05.360 --> 00:35:11.680 cannot see it. Why? Because what I see here as a  subtitle can actually be spoken by a screen reader  00:35:12.800 --> 00:35:17.040 and I show that to you now on those slides because obviously  00:35:17.040 --> 00:35:20.800 what the one thing I don't want you to say is “well, this is the MathCheck web  00:35:20.800 --> 00:35:23.680 page, of course you've got all kinds of fancy things in there.  00:35:23.680 --> 00:35:27.440 But can I do that as well. yes you can of course because  00:35:27.440 --> 00:35:31.920 these slides here are running at home on my computer. it's a website of course but  00:35:31.920 --> 00:35:35.040 MathCheck is somewhere in the background. so here's MathCheck I can  00:35:35.040 --> 00:35:38.400 see, I can, I can get to that formula I can  00:35:38.400 --> 00:35:41.520 see my original LaTex code which I put into the site  00:35:42.080 --> 00:35:48.560 this is the original later code and what I can now do is I can switch on a  00:35:48.560 --> 00:35:52.880 screen reader here on my machine and I can listen to what that formula has to  00:35:52.880 --> 00:35:56.560 say and let me do that here's my  00:35:57.120 --> 00:36:01.280 here, here are my chrome extensions so I’m, I’m working just for  00:36:01.280 --> 00:36:04.480 as background I’m working on a on a Linux box here  00:36:04.480 --> 00:36:08.400 the only screen I have installed is ChromeBox which runs in the in   00:36:08.400 --> 00:36:09.120 every chrome  00:36:09.920 --> 00:36:14.240 ChromeBox spoken feedback is ready and I hope you can hear that  00:36:16.240 --> 00:36:21.920 yes, yes, good let me go back to my x equals start fraction negative b plus  00:36:21.920 --> 00:36:24.800 or minus start root b squared minus four a c   00:36:24.800 --> 00:36:26.400 and root over 2 n fraction  00:36:27.520 --> 00:36:32.080 generic container generic container x equals start from  00:36:32.080 --> 00:36:36.000 sorry, let's, let's stop ChromeBox here for a second so you could already hear  00:36:36.000 --> 00:36:38.320 that it was speaking out the formula nicely  00:36:38.320 --> 00:36:42.800 right so we're speaking the formula but generally if you, if you listen to that  00:36:42.800 --> 00:36:45.840 formula even if you listen to it fast it's fairly long  00:36:45.840 --> 00:36:49.520 although this is a simple formula um so you might want to  00:36:49.520 --> 00:36:54.480 to step through it in more detail and so you can effectively start exploring this  00:36:54.480 --> 00:36:57.840 interactively x equals start fraction negative b plus  00:36:57.840 --> 00:37:01.760 or minus start root b squared minus four a c and root over two  00:37:01.760 --> 00:37:04.720 n fraction all right so now it, read it, read it  00:37:04.720 --> 00:37:08.320 again and we can now go into x equals  00:37:09.360 --> 00:37:12.480 start fraction negative b plus or minus star  00:37:12.480 --> 00:37:16.320 we can do this fraction, sorry, we can listen to it over and over again of  00:37:16.320 --> 00:37:18.640 course but I would like to point out a couple of  00:37:18.640 --> 00:37:20.960 things to you which you might not be familiar with  00:37:21.680 --> 00:37:25.600 so there's a couple of words in here which we would probably not  00:37:25.600 --> 00:37:29.760 use if you were to say write it on the blackboard and, and then and while  00:37:30.320 --> 00:37:33.680 writing it on blackboard you speak it out to your students you wouldn't say  00:37:33.680 --> 00:37:36.640 start fraction you wouldn't say start root or end root  00:37:36.640 --> 00:37:42.880 now these words are in there because imagine you cannot see what that formula  00:37:42.880 --> 00:37:45.600 is and you have to still get a feeling for  00:37:45.600 --> 00:37:48.560 the two-dimensional layout of that formula  00:37:48.560 --> 00:37:51.600 then it's important that you know which part  00:37:51.600 --> 00:37:56.640 is in the first in the fraction or which one is in the denominator which part is  00:37:56.640 --> 00:37:59.760 in the enumerator where does the square root  00:37:59.760 --> 00:38:03.840 start where does it end before you carry on  00:38:04.800 --> 00:38:08.400 and naturally this, this is, doesn't really sound very natural there's  00:38:08.400 --> 00:38:11.520 different ways you can pronounce this I’ll give you a different one you can  00:38:11.520 --> 00:38:15.760 switch the more natural way the fraction with numerator negative b  00:38:15.760 --> 00:38:20.320 plus or minus the square root of b squared minus 4ac and denominator 2r  00:38:20.960 --> 00:38:25.120 right, um, so this is a more natural way of reading it  00:38:25.120 --> 00:38:29.440 it's more natural however it's less disambiguating meaning that the more  00:38:29.440 --> 00:38:34.320 complex the formula becomes the harder will be, be for students to  00:38:34.320 --> 00:38:37.360 actually listen to it and know exactly where they  00:38:37.360 --> 00:38:43.760 are in the formula right so this is so as I said all this  00:38:43.760 --> 00:38:46.320 is being rendered with MathCheck in the background  00:38:46.960 --> 00:38:51.600 and the actual input is LaTex it also takes things like AsciiMath and MathML  00:38:52.160 --> 00:38:56.320 and it produces the output straight away in the browser  00:38:57.840 --> 00:39:02.480 and there's nothing really to do apart from putting your LaTex into it and one  00:39:02.480 --> 00:39:04.720 extra line into the HTML   00:39:04.720 --> 00:39:09.360 code which calls MathCheck from the web and that's it and things start rendering  00:39:09.360 --> 00:39:12.320 and the accessibility is built in so in many ways  00:39:12.320 --> 00:39:15.680 there is no excuse to have non-accessible math on,  00:39:15.680 --> 00:39:19.600 on, on web pages unfortunately they're still of that nature out there  00:39:20.320 --> 00:39:23.920 um let me switch off my screen here because otherwise it'll interfere  00:39:26.720 --> 00:39:33.120 we have two questions that I wanted to all right see if you could answer um one  00:39:33.120 --> 00:39:40.400 is whether students who frequently um look at math read aloud to them  00:39:40.400 --> 00:39:42.560 understand the way it's spoken by MathJax  00:39:44.000 --> 00:39:46.400 um right that that is a good question  00:39:48.240 --> 00:39:52.240 there's so, so what we implement we implement two particular grammars in  00:39:52.240 --> 00:39:56.400 there um one is called math speak  00:39:56.400 --> 00:39:59.600 so that's a, it's, it's a grammar that goes back to  00:40:00.320 --> 00:40:04.000 um Abraham Nemeth the person who also did uh  00:40:04.000 --> 00:40:07.760 math braille and he kind of put together that grammar as a  00:40:07.760 --> 00:40:11.440 standard grammar how things should be spoken it's of course enhanced because  00:40:11.440 --> 00:40:14.400 we do a lot more kind of semantic recognition if you like  00:40:14.400 --> 00:40:18.720 where we recognize more mathematical constructs which then can be spoken  00:40:18.720 --> 00:40:22.240 directly and the other one the, the second one I,  00:40:22.240 --> 00:40:26.640 I showed you is called clearSpeak which was developed a few years back  00:40:26.640 --> 00:40:31.840 maybe 10 years ago years ago by, by ETS, by the electronic testing service  00:40:31.840 --> 00:40:36.960 um so students are generally familiar with MathSpeak  00:40:36.960 --> 00:40:43.280 and please speak up to a certain level and as it's more natural and it's more  00:40:43.280 --> 00:40:46.480 in line with what teachers would say I assume they  00:40:46.480 --> 00:40:50.960 understand it quite well um I, I have to admit uh  00:40:50.960 --> 00:40:54.800 we have, we have done quite a bit of testing particularly recently  00:40:54.800 --> 00:40:59.680 we've done testing in, in l, I within a different project uh with students in  00:40:59.680 --> 00:41:02.160 India and there is a certain learning curve  00:41:03.680 --> 00:41:06.720 but once they have that learning curve in particular once they have the  00:41:06.720 --> 00:41:11.920 interaction as well available then it becomes quite uh so, so they  00:41:11.920 --> 00:41:14.960 become quite proficient at it was there a second question that's  00:41:14.960 --> 00:41:19.040 really helpful yeah uh second question is I’m getting a lot of  00:41:19.040 --> 00:41:23.600 reverb here I’m not sure why but I probably, ChromeBox is always  00:41:23.600 --> 00:41:30.480 making problems when just on and off. um this question asks whether there's a  00:41:30.480 --> 00:41:34.000 way to copy an equation from a page with MathJax  00:41:34.000 --> 00:41:37.200 in it and then paste the verbal description in another document  00:41:39.920 --> 00:41:43.520 of course there is but you wouldn't, you wouldn't really want to do that  00:41:45.920 --> 00:41:51.360 you, you can't do it here just like this um however  00:41:53.120 --> 00:41:56.240 that there are several ways to do it so so one way is you  00:41:56.240 --> 00:42:00.160 with on MathCheck we've got a page where you can get all in various uh  00:42:00.160 --> 00:42:03.760 locales you can describe verbal descriptions out for formulas  00:42:04.960 --> 00:42:09.040 another way and a much better way to do this is if you want to do that for an  00:42:09.040 --> 00:42:14.320 entire document we do have uh scripts and  00:42:14.320 --> 00:42:21.360 and, and uh programs that let you do this in bulk for in within your accessibility  00:42:21.360 --> 00:42:23.120 workflow I will discuss some   00:42:23.120 --> 00:42:24.720 workflows towards the end  00:42:25.440 --> 00:42:29.040 time permitting I, I’m mindful that the time is already  00:42:29.040 --> 00:42:33.760 foreign um but there is a number of workflows which you can use  00:42:33.760 --> 00:42:38.160 just to, to get speech out and, and to get speech integrated in  00:42:38.160 --> 00:42:41.680 into your, into your documents if you like. that sounds great.  00:42:41.680 --> 00:42:45.840 thank you. I would like to briefly, I’m mindful of the time, I'd  00:42:46.640 --> 00:42:50.160 what happens actually in the background there which produces the speech is not  00:42:50.160 --> 00:42:51.200 MathCheck it's  00:42:51.200 --> 00:42:54.960 my very own little system called Speech Rule Engine which I originally started  00:42:54.960 --> 00:42:59.360 implementing while I was at Google working on ChromeBox which is now a  00:42:59.360 --> 00:43:03.840 speech solution for quite a number of systems out there including EquatIO by  00:43:03.840 --> 00:43:07.920 Texthelp obviously MathCheck,  MathLive and, and a number of others  00:43:07.920 --> 00:43:11.600 and of course it still runs on your everyday ChromeBook  00:43:11.600 --> 00:43:15.920 in ChromeBox and what really happens there is  00:43:15.920 --> 00:43:21.600 um in order to produce the speech or in order to produce good speech let's put  00:43:21.600 --> 00:43:24.640 it this way it's not enough just to look at the  00:43:24.640 --> 00:43:28.640 actual representation of the mathematics as you would find it in LaTex or in  00:43:28.640 --> 00:43:34.560 MathML because that is just very, very flat, right? What you really want to be able to  00:43:34.560 --> 00:43:39.200 do is you want to understand where say where an equation is being split up what  00:43:39.200 --> 00:43:44.160 operators are in these things and then use that information and what happens is  00:43:44.160 --> 00:43:47.120 that there's a pattern recognition in the background of course and  00:43:47.120 --> 00:43:50.640 there's a semantic representation that is being created bottom up  00:43:50.640 --> 00:43:54.080 from, from the from the visual representation  00:43:54.080 --> 00:43:57.600 and I’ll just show you how that semantic representation looks like for a very  00:43:57.600 --> 00:44:02.000 simple equation like a x plus square plus b x plus c so the  00:44:02.000 --> 00:44:05.840 semantic representation is represented as a tree where you then get  00:44:05.840 --> 00:44:10.400 an equality of two sides on the zero on the right inside on the left-hand side  00:44:10.400 --> 00:44:13.440 you've got a sum with three summons and so on and so forth  00:44:14.000 --> 00:44:17.440 and this is in many ways uh how the actual  00:44:18.160 --> 00:44:22.880 representation then is done in the background and how it's being folded  00:44:22.880 --> 00:44:29.200 into, into the formula that is being displayed and thereby we  00:44:30.000 --> 00:44:33.840 we make use of what's called the separation of concerns for the web  00:44:33.840 --> 00:44:37.280 that you have parts which are responsible  00:44:37.280 --> 00:44:41.840 for purely for the display and the styling which is primarily HTML and CSS  00:44:42.400 --> 00:44:46.000 and then we have various parts which allow us to hide  00:44:46.000 --> 00:44:49.840 information in these structures without them actually being visible  00:44:50.400 --> 00:44:53.920 and some of them is for instance WAI-ARIA which is a hum which  00:44:53.920 --> 00:44:57.680 tends towards human oriented semantics uh which is  00:44:57.680 --> 00:45:01.840 for, for accessibility and there's various other ways how to do that and  00:45:01.840 --> 00:45:05.920 JavaScript in order to make this the whole thing interactive and, and  00:45:05.920 --> 00:45:10.640 start working and so in many ways this semantic tree  00:45:10.640 --> 00:45:15.760 I've just shown you on that page is folded thing of it as kind of an,  00:45:15.760 --> 00:45:20.640 a, a different view on the on the actual element that you see on that page  00:45:23.200 --> 00:45:26.960 and what that allows you once that rich semantic is in there what that also  00:45:26.960 --> 00:45:30.560 allows you is to go to, to produce other kind of  00:45:30.560 --> 00:45:32.880 formats so for instance here's another form,  00:45:32.880 --> 00:45:36.560 formula slightly more exciting this is a curl of vector field  00:45:37.120 --> 00:45:44.960 um which if when I start that now you get the same effect it takes a  00:45:44.960 --> 00:45:47.440 little while longer because that's not just  00:45:48.320 --> 00:45:51.040 speech because this now comes out as braille  00:45:51.040 --> 00:45:54.640 and that braille would now go to the student's uh braille display  00:45:54.640 --> 00:45:57.840 if they read it and if they have a braille display connected,  00:45:57.840 --> 00:46:01.120 right, you can still do the same thing. you can navigate the whole thing and you  00:46:01.120 --> 00:46:04.000 get the braille for every sub expression and so on and so forth  00:46:07.440 --> 00:46:11.520 and here the subtitle braille normally doesn't appear just done that so,  00:46:11.520 --> 00:46:14.560 so you can see what's happening uh if you  00:46:14.560 --> 00:46:17.520 if you were reading this on a braille display obviously you can  00:46:18.560 --> 00:46:21.920 all kinds of other formats you can generate out of that  00:46:21.920 --> 00:46:26.000 as soon as you have the interesting and enriched semantics in there  00:46:26.000 --> 00:46:28.400 you can actually generate all kinds of other um  00:46:30.400 --> 00:46:33.760 all kinds of other formats like braille, like Nemeth,  00:46:33.760 --> 00:46:40.320 like um computational formats etc. right I’m mindful of the time I briefly  00:46:40.320 --> 00:46:43.600 tell you about diagrams and then I’ll tell you a bit about these workflows.  00:46:46.400 --> 00:46:49.520 so what about diagrams? well diagrams are obviously even harder  00:46:50.240 --> 00:46:54.800 the problem is that Woocock tells us an alternative text is fine what for  00:46:54.800 --> 00:46:58.160 for any graphic or image um  00:46:58.160 --> 00:47:02.000 in practice when it comes to STEM that is obviously not true  00:47:02.000 --> 00:47:07.040 now if you have an alternative text normally has the length of 160  00:47:07.040 --> 00:47:10.480 characters I believe which is the size of a tweet  00:47:10.480 --> 00:47:14.560 and that's generally not enough to describe anything which is more complex  00:47:14.560 --> 00:47:19.040 than the uh but then say a parabola  00:47:21.360 --> 00:47:24.800 um so there's various ways how to do this  00:47:24.800 --> 00:47:28.160 I’m not going to go through, through those in detail now  00:47:28.160 --> 00:47:32.000 but the best way to do this is using a format called SVG  00:47:32.000 --> 00:47:35.120 SVG scalable vector graphic i've shown that to you early  00:47:35.120 --> 00:47:38.640 it scales it doesn't become grainy and you have all the  00:47:39.200 --> 00:47:42.720 advantage that you can embed lots of information  00:47:42.720 --> 00:47:46.160 in the background and now how that is being done  00:47:46.160 --> 00:47:51.120 is I demonstrate with a chem example from chemistry which we primarily  00:47:51.120 --> 00:47:58.400 working with now in my company um this here is just a a uh chemical  00:47:58.400 --> 00:48:03.680 diagram that is a bitmap image what happens to the bitmap image is that  00:48:03.680 --> 00:48:07.600 is being analyzed as image analysis run over  00:48:08.720 --> 00:48:12.640 one starts putting together a basic graph out of all these  00:48:12.640 --> 00:48:18.240 uh out of all the edges and then try to understand what those edges are all  00:48:18.240 --> 00:48:23.920 these top ledges or the single edges etc and then these there's a pattern there's  00:48:23.920 --> 00:48:27.840 pattern recognition algorithms in there which then will eventually put together  00:48:28.400 --> 00:48:34.400 a diagram of that of that molecule name naming components automatically  00:48:34.400 --> 00:48:36.880 using all kinds of web services out there  00:48:36.880 --> 00:48:40.960 and then generate a fully accessible diagram and I’ll show you the diagram  00:48:40.960 --> 00:48:44.560 here and and I’ll show you  00:48:46.080 --> 00:48:49.600 I’ll show you how that works bit similar to what we do in the mathematics  00:48:49.600 --> 00:48:54.160 um you can navigate this thing and again you get a subtitle which tells you what  00:48:54.160 --> 00:48:58.080 this is and you can see this is a  00:48:58.080 --> 00:49:01.840 this is an SVG I can zoom in it becomes nice and  00:49:02.480 --> 00:49:06.160 it scales not scales up nicely you can also  00:49:06.160 --> 00:49:09.200 do tricks like change the background for instance  00:49:09.200 --> 00:49:11.360 right background colors all kinds of different  00:49:12.240 --> 00:49:16.000 background colors and then you can start navigating this  00:49:16.000 --> 00:49:20.560 expression using the cursor keys in that case for instance I use go down  00:49:20.560 --> 00:49:24.640 and I’ll start zooming in onto the first major component which is a few string  00:49:24.640 --> 00:49:28.880 system and then I can stop go even further down  00:49:28.880 --> 00:49:32.400 this is a benzene ring and I can go over to the other ring and  00:49:32.400 --> 00:49:35.200 if I have no idea what the three oxolene ring is  00:49:35.200 --> 00:49:39.760 then I switch off the expert mode, I’ll get more information because there's  00:49:39.760 --> 00:49:44.240 additional knowledge in the graph and then that tells me this is a sub ring  00:49:44.240 --> 00:49:48.160 with five elements with oxygen at position one and there's  00:49:48.160 --> 00:49:50.560 a double bond between positions two and three  00:49:50.560 --> 00:49:54.080 three and four and again if I were to switch on the screen reader now you  00:49:54.080 --> 00:49:56.960 could actually listen to this and and  00:49:57.920 --> 00:50:02.160 it would, it would tell and, and the navigation would be described what you  00:50:02.160 --> 00:50:04.160 do right  00:50:06.960 --> 00:50:10.640 and now as I said all this is done with image analysis once you have  00:50:10.640 --> 00:50:16.560 more information to generate these, these images then it's even easier so for  00:50:16.560 --> 00:50:19.200 instance you can generate these things using  00:50:19.200 --> 00:50:23.520 chemical specifications that you get a regular chemical drawing programs like  00:50:23.520 --> 00:50:26.880 smiles or more or cml where you can pick them up on on  00:50:26.880 --> 00:50:31.360 Wikipedia or if you're working in different  00:50:31.360 --> 00:50:34.480 domains say you have want to do something with statistics  00:50:34.480 --> 00:50:37.920 once you have the statistical model of course then you already have that  00:50:37.920 --> 00:50:40.880 information all you then need to do is you need to  00:50:40.880 --> 00:50:43.360 have some means of embedding that information  00:50:44.080 --> 00:50:47.360 into your document so it can actually be made accessible  00:50:48.160 --> 00:50:52.160 and and there's friends one thing i've been working on with a colleague in new  00:50:52.160 --> 00:50:57.600 zealand is a package called braille r and that does effectively the  00:50:57.600 --> 00:51:01.200 same thing as we've shown I've showed you for the  00:51:01.200 --> 00:51:04.320 um chemical diagrams just for  00:51:05.440 --> 00:51:08.480 data visualization so for instance here's a box plot  00:51:09.040 --> 00:51:12.080 and I can now start navigating the box plot  00:51:12.080 --> 00:51:16.240 and I can go to the title I can look over the axis  00:51:16.800 --> 00:51:21.120 the y-axis and then I can start looking in the box plots. it tells me all the  00:51:21.120 --> 00:51:21.840 data points  00:51:22.480 --> 00:51:27.200 tells me minimum maximum and like this is kind of you can, you can dive into  00:51:27.200 --> 00:51:29.920 these you can dive into these and start  00:51:29.920 --> 00:51:34.080 walking around see what the quartiles are or you can move over to the next  00:51:34.080 --> 00:51:38.640 element and so on and so forth so you get a hierarchical view  00:51:38.640 --> 00:51:42.640 on these elements which allow you to actually experience  00:51:42.640 --> 00:51:46.480 um this, this diagram, the data visualization.  00:51:47.760 --> 00:51:51.600 and just to say that the person I did this with Jonathan Godfrey, he's a  00:51:51.600 --> 00:51:55.440 blind lecturer in statistics at, at Massey University in New Zealand,  00:51:56.080 --> 00:52:00.400 and for him why this was important was because he wanted to actually see  00:52:00.400 --> 00:52:00.960 whether  00:52:00.960 --> 00:52:03.840 the data visualization he gives his students  00:52:04.560 --> 00:52:08.480 were useful and, and, and he didn't always want to,  00:52:09.440 --> 00:52:13.360 if you work in a model you don't always want to, to emboss this onto an  00:52:13.360 --> 00:52:16.560 embosser then feel whether everything is there in the right place  00:52:16.560 --> 00:52:19.760 so at least you get an overview you can listen to things  00:52:19.760 --> 00:52:24.800 and, and see um whether whether they make sense before you then  00:52:24.800 --> 00:52:28.880 take the exact same data put it onto an embosser get an  00:52:28.880 --> 00:52:32.160 alternative medium and then have a tactile copy of it  00:52:33.760 --> 00:52:38.960 right, right now I've promised you that I wanted to give you some sample workflows  00:52:38.960 --> 00:52:41.520 how you can actually work with these things and how you can make things  00:52:41.520 --> 00:52:44.480 successful so one of the question earlier, earlier  00:52:44.480 --> 00:52:48.640 was, is there some way of getting um speech out directly and  00:52:48.640 --> 00:52:53.920 yes there is. so there's plenty of elements on our, on our, uh MathCheck  00:52:53.920 --> 00:52:57.120 website where you can take the math checks uh scripts directly  00:52:58.880 --> 00:53:01.920 and all of these scripts and all the workflows  00:53:01.920 --> 00:53:05.600 I’m talking about here one of the goals is first of all  00:53:06.320 --> 00:53:10.160 that you don't have to do things formula by formula but that you do things in  00:53:10.160 --> 00:53:14.080 bulk document by document  00:53:14.080 --> 00:53:17.920 that you do not need to change your authoring habits  00:53:17.920 --> 00:53:21.040 you do not need to move away from whatever if you  00:53:21.040 --> 00:53:25.040 if you're using LaTex you can use LaTex, if you're using Word you can use Word  00:53:25.760 --> 00:53:33.840 um, and also what is important is that they allow you  00:53:33.840 --> 00:53:36.720 to make your document future proof there's a lot  00:53:36.720 --> 00:53:39.920 of technology already out there to make things successful but  00:53:39.920 --> 00:53:44.800 if things come online more, more things come online on a daily basis  00:53:46.400 --> 00:53:49.920 what is important is that all the information that you have in your  00:53:49.920 --> 00:53:54.160 document is still preserved when it's being turned into an  00:53:54.160 --> 00:53:57.280 alternative format because there might be a system in a  00:53:57.280 --> 00:54:00.080 couple of years’ time which can exploit that  00:54:00.080 --> 00:54:03.200 might not exist yet but there might be one  00:54:03.200 --> 00:54:08.800 soon. right so here's a workflow which for instance allows you to just put  00:54:08.800 --> 00:54:12.640 LaTex, take LaTex sources and put it  00:54:12.640 --> 00:54:16.320 onto the web that's a simple workflow with Tech 4 HD.  00:54:16.320 --> 00:54:20.560 Tech 4 HD is a part of every standard tech Distribution,  00:54:21.120 --> 00:54:25.680 so if you have Tech-Life or if you have a MiKTeX or anything you normally have  00:54:25.680 --> 00:54:28.880 Tech 4 HD and its counterpart Make 4 HD,  00:54:30.400 --> 00:54:34.080 then all you need to do is a couple of very, very  00:54:34.080 --> 00:54:38.080 simple command line arguments on your file name  00:54:38.080 --> 00:54:40.400 and you would pretty much automatically  00:54:41.360 --> 00:54:45.280 get a fully accessible math document with and you can say whether you want to  00:54:45.280 --> 00:54:48.720 make it accessible with MathCheck so you just given MathCheck as an extra  00:54:48.720 --> 00:54:49.280 option  00:54:49.280 --> 00:54:53.280 it automatically includes MathCheck from the web and it'll run  00:54:53.280 --> 00:54:56.720 and will be accessible. obviously when i say  00:54:57.760 --> 00:55:01.200 all of it will like be accessible, there will always be some caveats, there will  00:55:01.200 --> 00:55:04.560 always be, you know, there will always be certain macro packages that won't  00:55:04.560 --> 00:55:08.080 understand and so on and so forth but that's more about authoring  00:55:08.880 --> 00:55:12.400 later rather than about, about the conversion itself  00:55:14.640 --> 00:55:18.160 you can do something fairly similar with a package called Pandoc.  00:55:18.160 --> 00:55:22.400 Pandoc is a universal document converter and I’ll just show you briefly um  00:55:22.400 --> 00:55:25.760 what it can do let's scroll down here there's the  00:55:25.760 --> 00:55:32.000 the whole the whole graph of all the input formats Pandoc can use  00:55:32.000 --> 00:55:34.880 and all the output formats Pandoc can review  00:55:35.520 --> 00:55:39.520 so whenever you need to transform from one format into the other  00:55:40.160 --> 00:55:42.640 Pandoc should be your first point of call.  00:55:43.680 --> 00:55:47.840 what it can also do is similar to Tech 4 HD. it can take  00:55:48.400 --> 00:55:51.280 but more in many ways it can take other  00:55:51.280 --> 00:55:54.320 sources like Word markdown etc.  00:55:55.360 --> 00:56:00.000 and then output HTML for instance but also other things so you can translate  00:56:00.000 --> 00:56:03.520 LaTex into Word you can translate markdown in the LaTex  00:56:03.520 --> 00:56:10.080 and whatever you like. um you can include MathCheck directly uh you can  00:56:10.080 --> 00:56:14.240 also tell it not to produce, do anything smart with the tech and just  00:56:14.240 --> 00:56:17.920 let MathCheck to all these jobs. um  00:56:24.720 --> 00:56:29.360 um right and these in many ways are here's the idea of these workflows is  00:56:29.360 --> 00:56:35.280 that you translate from something um they never launched into something, into  00:56:35.280 --> 00:56:39.360 another benevolent um, um, format I would call those  00:56:40.800 --> 00:56:44.080 there's malevolent formats out there and those are the, what  00:56:44.080 --> 00:56:48.720 print formats and one of the old print formats of were for instance Postscript  00:56:48.720 --> 00:56:50.960 but the new print format of course is still  00:56:50.960 --> 00:56:54.560 and still is PDF and the problem with PDF is  00:56:54.560 --> 00:56:57.840 that once things are in PDF it's really a nightmare,  00:56:58.640 --> 00:57:03.360 or it can be a nightmare, and it's difficult to generate accessible  00:57:04.080 --> 00:57:09.520 PDF as it is right now. Um, there are important things which you  00:57:09.520 --> 00:57:12.720 often get wrong or which the generators often get wrong that's  00:57:12.720 --> 00:57:16.640 things like reading order, alt-texts, header nesting,  00:57:16.640 --> 00:57:22.960 and you often have to work a lot with Acrobat Pro, Adobe Acrobat Pro in order to get  00:57:22.960 --> 00:57:27.760 these things right manually. it's particularly bad if you're coming  00:57:27.760 --> 00:57:32.320 from LaTex you have to do a lot of work. LaTex is  00:57:32.320 --> 00:57:36.880 really bad in producing accessible PDF, Word is better so if you  00:57:37.680 --> 00:57:41.520 if you need to translate LaTex and access and PDF it's sometimes better to  00:57:41.520 --> 00:57:47.680 go via word by using, using Pandoc, um often what happens is you make things  00:57:47.680 --> 00:57:50.960 accessible nicely in your Acrobat Reader or Acrobat Pro  00:57:50.960 --> 00:57:55.120 and then you then you find that typo and you have to start all over again.  00:57:55.120 --> 00:57:59.360 it's, it's really not nice it's getting even worse if you only have PDF as a  00:57:59.360 --> 00:58:01.920 source. so there's quite a bit of work has been  00:58:01.920 --> 00:58:05.840 done uh on that problem. turns out it's actually a  00:58:05.840 --> 00:58:09.600 surprisingly difficult problem. so say you have a PDF document that has been  00:58:09.600 --> 00:58:13.120 originally generated with LaTex but all the sources are gone.  00:58:15.680 --> 00:58:19.520 um then what you're trying to do is you're trying to get the information  00:58:19.520 --> 00:58:23.280 that's in the PDF back out and that is quite a nasty  00:58:23.280 --> 00:58:28.560 pattern recognition problem actually. in about 2006 I had a PhD student who  00:58:28.560 --> 00:58:32.000 was supposed to do as a warm-up exercise for his PhD,  00:58:32.800 --> 00:58:37.520 extraction of math elements from PDF. eventually it turned out  00:58:37.520 --> 00:58:41.120 to be his entire PhD and the problem is still not fully solved.  00:58:41.920 --> 00:58:46.400 we brought a system out which was called Maxtract at the time 2008, which we're  00:58:46.400 --> 00:58:48.720 currently reworking. I’m really working with the group in  00:58:48.720 --> 00:58:53.040 India at, at Delhi. um other people have re-implemented some  00:58:53.040 --> 00:58:55.440 of our approaches particularly the Infty system  00:58:55.440 --> 00:59:02.160 which generally also does other math OCR. um what I can say is the standard is  00:59:02.160 --> 00:59:05.600 currently improving but it's not quite there yet so with any  00:59:05.600 --> 00:59:08.240 luck at some point in the future you might be able to  00:59:08.240 --> 00:59:10.800 make these inaccessible PDFs accessible again  00:59:11.360 --> 00:59:17.840 um but it might still take a while, will never be perfect.  00:59:19.120 --> 00:59:25.040 um let me have a quick check at the time yes so another  00:59:25.040 --> 00:59:28.880 alternative um workflow is which we're  00:59:29.680 --> 00:59:36.560 currently trying to perfect and finalize with uh is trying to get from LaTex in  00:59:36.560 --> 00:59:41.920 particular in PreTeXt format um to emboss braille so where you can  00:59:41.920 --> 00:59:45.840 effectively have a textbook in PreTeXt which PreTeXt is just  00:59:46.880 --> 00:59:51.280 LaTex with some XML markup so it takes the good parts of LaTex think about it  00:59:51.280 --> 00:59:55.840 this way the math markup and replaces the bad parts uh, the, the  00:59:56.560 --> 00:59:59.680 document structure markup, uh with pro,  00:59:59.680 --> 01:00:04.400 with XML and from that we're  using MathCheck and SRE  01:00:04.400 --> 01:00:07.520 and, and some other libraries in order to get  01:00:07.520 --> 01:00:11.920 it straight into Nemeth output we're working on that with the NFB  01:00:11.920 --> 01:00:15.120 and we have grant from the from the American Action Fund  01:00:15.120 --> 01:00:19.600 to do that and uh it's currently going pretty well.  01:00:23.040 --> 01:00:26.000 I’m mindful of the time so I’ll skip the graphics part.  01:00:26.640 --> 01:00:29.760 the graphics is quite straightforward. I've shown you some examples  01:00:29.760 --> 01:00:36.480 um please on the slides there's various links and, and, and you can certainly  01:00:36.480 --> 01:00:43.360 um have a look at that. what else did I want to say? well um one  01:00:43.360 --> 01:00:46.720 thing is as I said before please  01:00:47.440 --> 01:00:50.960 let nobody tell you that you have to change your authoring workflow  01:00:51.520 --> 01:00:56.240 um because this is just, you know, this impedes creativity in my opinion.  01:00:56.800 --> 01:01:01.840 but be mindful when you author that you might be authoring for something that  01:01:01.840 --> 01:01:08.960 goes beyond just that one PDF format. you generate um your, your document for.  01:01:08.960 --> 01:01:14.720 it goes beyond that single, uh journal you're submitting your paper to and  01:01:14.720 --> 01:01:15.360 therefore  01:01:15.360 --> 01:01:19.280 don't, stop thinking print only particularly, that particularly means  01:01:19.280 --> 01:01:23.120 stop thinking about manipulating the white space.  01:01:23.120 --> 01:01:26.880 in particular when you're talking math formulas, do not put white space  01:01:26.880 --> 01:01:30.240 manipulations into your math formulas, the slash,  01:01:30.240 --> 01:01:33.360 exclamation mark, and all these nasty bits  01:01:33.360 --> 01:01:36.560 is really not good. it's really, really, really bad.  01:01:36.560 --> 01:01:42.400 for you can imagine if we go out from single pages to this reflowing,  01:01:42.400 --> 01:01:46.400 nicely reflowing document you do not want any hard line breaks,  01:01:46.400 --> 01:01:49.440 hard page breaks or anything of that nature because that  01:01:49.440 --> 01:01:51.840 ultimately that has to be filtered out somehow.  01:01:54.320 --> 01:01:57.840 um there's a lot more tips I would have on math authoring.  01:01:57.840 --> 01:02:04.640 um just a few that I would like to, if you write math that should  01:02:04.640 --> 01:02:07.120 eventually be converted into different format  01:02:07.120 --> 01:02:10.240 think about not using all the, your homegrown macros.  01:02:10.880 --> 01:02:14.720 think about using more standard environments  01:02:14.720 --> 01:02:19.360 as they are for list figures etc. don't use custom elements like  01:02:19.360 --> 01:02:25.920 list items, specialist list items etc. and even if you're doing slides say with  01:02:25.920 --> 01:02:32.160 Beamer, Pandoc does a great job translating those into web formats for instance  01:02:32.160 --> 01:02:35.120 but it has a hard time when you're using non-standard  01:02:36.000 --> 01:02:40.160 elements. other things from a MathCheck point of view  01:02:40.800 --> 01:02:45.920 um try to write clean math, don't do too much nesting of text and math.  01:02:46.480 --> 01:02:48.480 from an accessibility point of view,.  01:02:50.800 --> 01:02:55.040 no punctuation after your math elements please, please, please if you write your  01:02:55.040 --> 01:02:57.360 formulas why do you have to put a dot behind it  01:02:58.480 --> 01:03:02.160 because what happens when, when you try to make that accessible into speech  01:03:02.160 --> 01:03:05.440 it becomes two elements: it's the formula and then the dot.  01:03:05.440 --> 01:03:08.640 right so the first bit of navigation will go into the either the   01:03:08.640 --> 01:03:09.520 formula or the dot  01:03:10.080 --> 01:03:13.760 and then you know and, and only then you get to the real math  01:03:13.760 --> 01:03:17.840 and that is really, really you know, it's effectively a layer you put  01:03:17.840 --> 01:03:21.920 on, on top of your math which shouldn't be there and prevents accessibility.  01:03:24.960 --> 01:03:29.440 well finally once you're doing all these things  01:03:29.440 --> 01:03:33.360 or once you're trying to do all these things, is there some way to easily  01:03:33.360 --> 01:03:36.320 check whether what you've done is really accessible?  01:03:37.120 --> 01:03:41.120 Yes, there is of course. first thing is always try some zooming:  01:03:41.760 --> 01:03:45.040 does it work on all zoom levels so here's my,.  01:03:45.040 --> 01:03:48.080 I’m sure they don't work well, don't really work all that well  01:03:48.960 --> 01:03:52.720 um so here we see that's not good  01:03:57.120 --> 01:04:01.440 now next Pro Question is always what does it work on high contrast,  01:04:01.440 --> 01:04:05.600 high contrast is easy to check there's a there's a Chrome extension  01:04:05.600 --> 01:04:10.240 it's Shift F11, yep works on high contrast.   01:04:10.240 --> 01:04:11.840 I can see pretty much everything.  01:04:12.800 --> 01:04:16.800 next question is does it work for, for keyboard interaction, can you get to  01:04:16.800 --> 01:04:22.480 every link that's in the page? um I’m not good, well I can’t that here  01:04:22.480 --> 01:04:26.160 so here I am. well and that's  01:04:26.160 --> 01:04:29.360 um and I can get to all the links, yes, fine.  01:04:30.880 --> 01:04:35.120 other things you can try yourself, well you can try form factors and I’ll show  01:04:35.120 --> 01:04:38.160 that to you, you can switch on screen readers. if you're working on  01:04:38.160 --> 01:04:41.760 Apple voiceover is Command F5 and it Works.  01:04:42.560 --> 01:04:45.040 Windows you've got Narrators-Windows- Enter.  01:04:45.920 --> 01:04:50.560 you can easily, NVDA is free, JAWS is free for 40 minutes,  01:04:50.560 --> 01:04:53.840 easy to test. I use ChromeBox and Chrome, it's  01:04:53.840 --> 01:04:57.520 also nice. and what you can always do is, you can switch on  01:04:58.640 --> 01:05:03.200 the, the, the developers console. it's there, there's a, it's a Chrome.  01:05:03.200 --> 01:05:07.200 I want to see my slides on different form factor, there's a different form  01:05:07.200 --> 01:05:08.720 factor right, what is that?  01:05:09.440 --> 01:05:13.360 um oops, that always interferes a bit with my,  01:05:14.160 --> 01:05:16.960 here's a different form factor -let's get rid of this a bit-um  01:05:17.760 --> 01:05:22.640 so uh what, what do I want, I want Galaxy S5 so here  01:05:22.640 --> 01:05:26.960 this is how my slides look on Galaxy S5 this is how they look on an iPad  01:05:27.520 --> 01:05:33.120 um and then the final little tool I've already brought up here.  01:05:33.120 --> 01:05:36.880 there's an accessibility checker here you can generate a report with the  01:05:36.880 --> 01:05:40.000 Lighthouse checker takes a while, um  01:05:41.760 --> 01:05:45.600 but once it's there it will tell you what the problems are,  01:05:45.600 --> 01:05:49.040 particularly the accessibility problems performance problems, all kinds of other  01:05:49.040 --> 01:05:50.080 little things as well.  01:05:51.120 --> 01:05:56.560 um there you go, at least mine slides, so they should be 100 accessible,  01:05:57.520 --> 01:06:01.520 not fully best practice performance is slightly less than it could be  01:06:01.520 --> 01:06:05.440 but you know it gives you a good feeling but now we're nearly there  01:06:05.440 --> 01:06:08.560 and we're on slide one which is obviously not where we want to be.  01:06:08.560 --> 01:06:15.520 um right so I think I should conclude. um um in the meantime let me try to get to  01:06:15.520 --> 01:06:20.000 the final slides. this is,  01:06:23.280 --> 01:06:26.880 and then I think I could take questions if. oops, I saw it.  01:06:30.560 --> 01:06:34.320 Conclusions? well I think my basic conclusions are  01:06:35.920 --> 01:06:39.520 things, accessible STEM is possible  01:06:39.520 --> 01:06:43.920 and it's the right time to do it now. you have to move your stuff online or, or  01:06:43.920 --> 01:06:47.120 into electronic format. anyway it's important that  01:06:47.760 --> 01:06:51.040 you use that opportunity rather than waste it.  01:06:52.000 --> 01:06:55.760 what's important is that you get, don't change your workflow,  01:06:55.760 --> 01:07:00.560 try to do things in bulk, and be aware of some of the do's and don'ts,  01:07:03.600 --> 01:07:08.320 and stay future-ready. meaning do not throw away your sources. try to integrate  01:07:08.320 --> 01:07:12.400 your sources as much as possible if you want to dive any deeper into the subject  01:07:13.120 --> 01:07:16.320 I have given plenty of talks of that nature it's  01:07:16.320 --> 01:07:19.360 confidence is like Accessing Higher Ground, CSUN,  01:07:19.360 --> 01:07:24.480 Empower. we had an AMS workshop on, in January. We’ll probably put  01:07:24.480 --> 01:07:31.840 up, uh put on another one at some point uh this year. um it's all about getting  01:07:31.840 --> 01:07:35.920 mathematic, mathematics online. um they, we  01:07:35.920 --> 01:07:39.520 might do a workshop at the joint math meetings as well, together with Peter  01:07:39.520 --> 01:07:43.760 Graudsberger And, and um Tom Byrne from the AMS.  01:07:44.640 --> 01:07:49.280 um there's the MathCheck documentation out there. we also offer bespoke training  01:07:49.280 --> 01:07:53.520 and seminars if you want us to, uh from MathCheck as well as with, with  01:07:53.520 --> 01:07:57.280 Peter Grouse playing with other people so there's plenty of  01:07:57.280 --> 01:08:00.240 things out there. I’m sorry for overrunning.  01:08:05.280 --> 01:08:10.720 Hello? um, you're fine I’m going to mute you while I’m talking.  01:08:12.640 --> 01:08:16.720 uh let's see okay, that's better, thank you. the echo is gone.  01:08:16.720 --> 01:08:21.840 um thank you that, this is very interesting. we have a number of  01:08:21.840 --> 01:08:25.840 questions from uh, from the panelists who can't ask  01:08:25.840 --> 01:08:29.520 questions um, and a number of questions in the Q and A  01:08:30.160 --> 01:08:32.560 and it was, your timing was just perfect so  01:08:33.200 --> 01:08:38.400 um I want to invite a couple colleagues Christa Miller and  01:08:40.080 --> 01:08:46.000 Lizette Zeitzman to unmute and to turn on cameras. um   01:08:46.000 --> 01:08:46.880 Christa Miller is  01:08:47.520 --> 01:08:51.840 the, I’ll let you actually introduce yourself.  01:08:54.080 --> 01:09:00.240 um sure I’m the Director of inclusive media design at Virginia Tech which  01:09:00.240 --> 01:09:03.680 basically means if a student has a need for  01:09:04.560 --> 01:09:08.000 alternative text I try to help provide it and  01:09:08.000 --> 01:09:12.560 that I teach a lot of faculty how to improve their authoring techniques for  01:09:12.560 --> 01:09:17.120 accessibility. and Lizette would you like to also  01:09:17.120 --> 01:09:20.640 introduce yourself? yeah it's, good afternoon my name is  01:09:20.640 --> 01:09:23.280 Lizette Zietzmann. I’m in the Department of Mathematics  01:09:24.000 --> 01:09:27.280 and I've been working with Anita for the past few years  01:09:27.280 --> 01:09:31.760 on developing open education resources for mathematics.  01:09:31.760 --> 01:09:36.160 as you know we have huge classes, thousand, two thousand students and  01:09:36.160 --> 01:09:40.400 who have to buy textbooks, so we hope that there are some initiatives  01:09:40.400 --> 01:09:44.560 um, that we can incorporate, and this was extremely helpful. thank you very much.  01:09:46.720 --> 01:09:52.400 Thank you both. uh Kayla McNabb is also Here. I, I recruited these three experts because  01:09:52.400 --> 01:09:55.840 I wanted uh people who know the subject material uh  01:09:56.720 --> 01:10:00.160 to um have a chance at um  01:10:00.160 --> 01:10:04.240 selecting some of the questions that we wrote before the panel,  01:10:04.240 --> 01:10:07.760 um as well as those that have come in through the Q and A so  01:10:07.760 --> 01:10:10.560 um Kayla, would you also introduce yourself?  01:10:12.000 --> 01:10:16.480 Certainly. thank you. uh I’m Kayla McNabb. I work in the University Libraries here  01:10:16.480 --> 01:10:19.120 at Virginia Tech. I’m Head of Instructional Content  01:10:19.120 --> 01:10:24.320 and Design. so we primarily focus on creating online content to support  01:10:25.120 --> 01:10:27.120 learning initiatives for the library,  01:10:27.120 --> 01:10:30.240 particularly digital literacy and information literacy initiatives.  01:10:33.040 --> 01:10:36.560 thank you. so we have a few questions in the chat  01:10:37.120 --> 01:10:40.800 and we have a few questions as, um as panelists.  01:10:42.080 --> 01:10:49.120 one of the, one of the questions that um I think Lizette was asking  01:10:49.840 --> 01:10:55.520 earlier is about operating systems and software. Lizette, do you want to  01:10:55.520 --> 01:10:56.720 jump into  01:10:56.720 --> 01:11:00.160 I’ll post your question also in the in the chat.  01:11:01.280 --> 01:11:03.920 um so everyone can see it I can't put it in today.  01:11:05.760 --> 01:11:11.840 um you mentioned future-proof material and you talked about robustness.  01:11:12.880 --> 01:11:17.360 and how, how much control do you have, so we,  01:11:17.360 --> 01:11:22.080 recently had a situation, we used a canvas testing system and the day of the  01:11:22.080 --> 01:11:22.880 exam  01:11:22.880 --> 01:11:28.720 the minus signs didn't show up. turned out it was on the Chromium browsers. they  01:11:28.720 --> 01:11:31.280 had a hardware acceleration setting change.  01:11:31.920 --> 01:11:34.960 but, so thousands of students took the exam that day.  01:11:35.760 --> 01:11:39.840 so how do you keep current? how do you know when it's going to be an issue?  01:11:41.040 --> 01:11:44.480 um well  01:11:44.480 --> 01:11:47.920 that, that is a good question and it's a nasty question as well, because that,  01:11:47.920 --> 01:11:50.880 that's something that's very much out of our control often.  01:11:51.680 --> 01:11:56.480 so, so okay I’ll, I’ll answer this from a MathCheck perspective.  01:11:58.000 --> 01:12:01.920 so the uh the idea of MathCheck is to provide a, a library  01:12:01.920 --> 01:12:06.320 which uh runs in every browser on every operating system and it renders the  01:12:06.320 --> 01:12:09.920 mathematics. however there are certain glitches which  01:12:09.920 --> 01:12:13.680 happen during the rendering, when the browsers, particularly when they change  01:12:13.680 --> 01:12:18.480 from one development mode to the other and, and so on and so forth. so we get  01:12:18.480 --> 01:12:20.800 quite a lot of issues which say things like  01:12:20.800 --> 01:12:23.840 okay, well this just disappeared in this zoom, uh  01:12:23.840 --> 01:12:27.360 zoom size on this screen on Chrome for instance. Chrome is  01:12:27.360 --> 01:12:29.920 very, because they update in the background all the time.  01:12:30.800 --> 01:12:34.560 and then there's that's one thing, like one thing you can always say is well  01:12:34.560 --> 01:12:35.440 wait a day  01:12:35.440 --> 01:12:38.720 it might change. you know, it might change for the better. that doesn't help, of course,  01:12:38.720 --> 01:12:42.240 if you're running that exam right now and I appreciate this.  01:12:42.240 --> 01:12:45.120 one thing that often helps with some of these things, is  01:12:45.920 --> 01:12:51.040 changing things like zoom factor if you, if you're running things like  01:12:51.040 --> 01:12:53.680 with MathCheck, you can change the renderer.  01:12:54.320 --> 01:12:58.640 if it happens in HTML mode, we have other renderers like SVG  01:12:58.640 --> 01:13:04.080 that might help, and if you're working with something and I, I know that  01:13:04.080 --> 01:13:06.800 is very frustrating, if you're working with something like  01:13:06.800 --> 01:13:11.760 Canvas or Blackboard unless you have control over it or unless you have  01:13:11.760 --> 01:13:15.120 control of the people who control it on your campus,  01:13:15.120 --> 01:13:18.160 you're often stuck with whatever they provide you with, with  01:13:18.720 --> 01:13:22.640 so often they don't switch on a single dollar sign  01:13:22.640 --> 01:13:26.800 math, um inline math note. right because that's a, that's an option  01:13:27.600 --> 01:13:31.440 uh that's an, uh that's a configuration option people don't often, don't know  01:13:31.440 --> 01:13:34.880 about it, or because you know that they are not technically  01:13:34.880 --> 01:13:37.280 knowledgeable enough to know what people need for math. And,  01:13:37.280 --> 01:13:41.440 and these are things you have to, you know you, you have to be aware of and  01:13:41.440 --> 01:13:47.200 then when you can tell people explicitly this is what I need and want. I can do it  01:13:47.200 --> 01:13:51.920 for you often. Right. so this is the long answer to that  01:13:51.920 --> 01:13:55.600 but, but the minus sign, I’m sorry that's beyond my control personally.  01:13:57.360 --> 01:14:01.600 um Volker there's, there's a few people who have kind of a follow-up question  01:14:01.600 --> 01:14:06.240 to the learning management system piece. so Canvas in particular has  01:14:06.240 --> 01:14:10.400 the HTML embed option and several people want to know  01:14:10.400 --> 01:14:13.680 if they can use these techniques and just  01:14:13.680 --> 01:14:18.880 paste it into the HTML and have it work the way you demonstrated it,  01:14:19.760 --> 01:14:22.880 or if there's another way to integrate MathJax into  01:14:23.440 --> 01:14:26.160 the Virginia Tech instance of Canvas for example.  01:14:26.720 --> 01:14:30.560 so the first thing is you should have a look at a, check with your, with your  01:14:30.560 --> 01:14:33.680 technical services whether they have MathCheck in there.  01:14:35.200 --> 01:14:40.080 most universities do. as I said, I mean I can only talk about my own experience  01:14:40.080 --> 01:14:45.440 from Canvas and at Birmingham and there is MathCheck in  01:14:45.440 --> 01:14:48.080 there. it's a very old version um  01:14:49.760 --> 01:14:54.960 and, and it's a fairly old version. it's not very old, but it's fairly old.  01:14:54.960 --> 01:14:59.120 uh inline math isn't switched on properly, like this is pretty but, but it  01:14:59.120 --> 01:15:02.720 works so you can, you can get your display math expression in  01:15:02.720 --> 01:15:07.440 there with two dollar signs at least and, and it should work. and if it works,  01:15:07.440 --> 01:15:10.800 it might be the same at Virginia Tech. obviously I wouldn't  01:15:10.800 --> 01:15:13.680 know. I probably don't even have access to your Canvas.  01:15:14.320 --> 01:15:18.800 um all I know is that at Birmingham they asked me whether I can update it for  01:15:18.800 --> 01:15:21.120 them and I told them yes I can send you a quote.  01:15:21.680 --> 01:15:24.480 I haven't heard back from them since um But,  01:15:27.440 --> 01:15:31.120 but yet but it's something for your technical staff people to, you know, if  01:15:31.120 --> 01:15:37.840 you know what you want, tell them and I’m sure they can do it.  01:15:39.680 --> 01:15:44.960 great okay. uh so there are a number of other questions, um  01:15:46.000 --> 01:15:51.200 that are kind of interesting. so uh one that everybody's interested in is, would  01:15:51.200 --> 01:15:54.000 you share a link for your slides? Um.  01:15:58.000 --> 01:16:01.440 okay yes yeah okay of course. I mean that's  01:16:01.440 --> 01:16:04.640 so, so the slide, just to talk a bit about my slides.  01:16:04.640 --> 01:16:09.040 that is a single file. with one exception, the MathCheck is being loaded in  01:16:09.040 --> 01:16:12.720 through the, uh through the web and it's a single  01:16:12.720 --> 01:16:16.960 file. everything is embedded including all the accessible elements  01:16:18.480 --> 01:16:21.920 and this is pretty much what I normally advocate. you know it's,  01:16:21.920 --> 01:16:26.560 it’s everything can be done in, in a single document that can be  01:16:26.560 --> 01:16:34.240 accessible all the, all the STEM in there. Sure, thank you. uh so another question  01:16:34.240 --> 01:16:39.120 um Jonathan Ports is asking if there should be a style file for  01:16:39.120 --> 01:16:44.560 LaTex which makes accessible PDFs? um there used to be one um  01:16:45.360 --> 01:16:48.800 it seems unsupported. do you know the history and do you know  01:16:49.440 --> 01:16:52.320 do you know any, anything about where that might be going?  01:16:54.800 --> 01:17:02.400 and yes, I do know a bit about it; there's been, it's, so this accessibility  01:17:02.400 --> 01:17:05.600 style it was never very good. I understand  01:17:05.600 --> 01:17:12.240 and it's, it's no longer supported, yes. Um there was a few years back, um,  01:17:13.440 --> 01:17:16.720 we tried to get something off the ground, so  01:17:16.720 --> 01:17:19.680 we actually also had some funding promised  01:17:20.560 --> 01:17:26.080 to make the whole process of LaTex turnout an  01:17:28.400 --> 01:17:31.440 accessible PDF. the problem with LaTex is  01:17:32.080 --> 01:17:38.720 that it, all this has to go through all the deep, really deep inner  01:17:38.720 --> 01:17:42.800 uh of the of the system in order to produce accessible PDFs.  01:17:42.800 --> 01:17:46.720 meaning that all the, most of these current style files have to be  01:17:46.720 --> 01:17:51.440 changed. so I don't um, it's, I don't really know all that much about it but  01:17:51.440 --> 01:17:55.600 but I trust the experts which, who looked into it at the time, so that was Ross  01:17:55.600 --> 01:17:58.240 Moore and he's, I think, he's in Canberra somewhere  01:17:58.240 --> 01:18:01.520 in Australia, who's the absolute expert on these things.  01:18:01.520 --> 01:18:06.640 um who can for instance, who's, I think, he's written a style file that at least  01:18:07.200 --> 01:18:12.640 in, I think, puts MathML for certain math equations into PDF  01:18:13.200 --> 01:18:18.160 which helps a bit um but, but Ross effectively said, well he's going to  01:18:18.160 --> 01:18:22.080 leave his hands off that because it is a can of worms he cannot open.  01:18:22.960 --> 01:18:26.080 And, which is very sad because you know in many ways  01:18:26.080 --> 01:18:29.840 everything should be there and in terms of structure everything is so much  01:18:29.840 --> 01:18:32.240 more neatly structured than, say, in Word  01:18:33.120 --> 01:18:36.000 but the conversion is just not that great.  01:18:37.360 --> 01:18:42.160 and so my advice is always, well, use Pandoc in order to go to a different format and  01:18:42.160 --> 01:18:48.880 go then from there into accessible PDF that's easier. we also have a Pandoc  01:18:48.880 --> 01:18:52.160 question: so if someone is using Pandoc and wants  01:18:52.160 --> 01:18:58.480 to HTML with the expressions in MathJacx can they tell Pandoc what to start, what  01:18:58.480 --> 01:19:02.960 start and end codes they would need? um  01:19:06.800 --> 01:19:12.720 right. so you can tell Pandoc, primarily you can tell Pandoc,  01:19:12.720 --> 01:19:15.520 first of all, to include MathCheck directly,  01:19:15.520 --> 01:19:22.160 and making the standalone file which generally leaves your LaTeX intact.  01:19:22.160 --> 01:19:25.200 and you can also, I think there's an option,  01:19:25.200 --> 01:19:28.320 which is called Rawtech it's on the slides.  01:19:28.320 --> 01:19:30.000 I think there's an option in Rawtech which  01:19:30.000 --> 01:19:35.840 means it will not touch any of the tech whatsoever.  01:19:36.400 --> 01:19:39.680 uh which means that it comes, yes, it's dash, dash, Raw  01:19:39.680 --> 01:19:44.400 dash, tech. and then it comes straight into MathCheck  01:19:44.400 --> 01:19:47.440 and, and we can do whatever, whatever we like with it  01:19:47.440 --> 01:19:52.480 or whatever we can. so what, what these ah converters often try to do  01:19:52.480 --> 01:19:56.000 is that they try to untangle some of the macros,  01:19:56.960 --> 01:20:00.960 and um depending on how much they are aware of style files,  01:20:01.760 --> 01:20:04.800 and, and that often, that sometimes goes wrong.  01:20:04.800 --> 01:20:09.200 Tech 4 HD is better for that than Pandoc.  01:20:15.200 --> 01:20:19.840 um okay, so we do have a note from one of our  01:20:19.840 --> 01:20:24.560 panelists that, um we'll follow up with the Canvas team about  01:20:24.560 --> 01:20:28.960 installing MathJax and that there's one, one workaround that  01:20:28.960 --> 01:20:35.280 is available in the meantime. yeah so um I would, I’m puzzling about  01:20:35.280 --> 01:20:37.360 this one. I’ll touch base with our Canvas team  01:20:37.920 --> 01:20:41.200 later this week, find out more about the MathJax but um  01:20:41.200 --> 01:20:47.120 in the meantime, we did a license and install a tool that's called  01:20:47.120 --> 01:20:50.640 equatIO. let me type the spelling in the chat for  01:20:50.640 --> 01:20:55.040 you. and, and really it's a MathML solution, so  01:20:55.840 --> 01:20:58.560 bear in mind that the things that Volker shared about that.  01:20:59.440 --> 01:21:02.640 it looks like a blue diamond, though, and you can  01:21:02.640 --> 01:21:08.800 access it directly from the, the editor inside Canvas whether it's on a page or  01:21:08.800 --> 01:21:12.800 a quiz. any of the above. and when you  01:21:12.800 --> 01:21:18.320 open it it actually has a LaTex option so that you can directly paste your  01:21:18.320 --> 01:21:21.040 LaTex code in there and you don't have to, like,  01:21:21.040 --> 01:21:26.000 recreate the equation manually or anything like that. so for  01:21:26.000 --> 01:21:29.120 for the LaTex lovers out there, that may produce a,   01:21:30.160 --> 01:21:34.960 and Lizette in particular, it might avoid the minus sign disappearing,  01:21:34.960 --> 01:21:38.400 um because it does create that, that MathML  01:21:39.280 --> 01:21:41.840 image so that's an option in the meantime.  01:21:44.160 --> 01:21:52.000 yes great, thank you for that equation. sorry do you want to say  01:21:52.000 --> 01:21:57.040 to equatIO? it is so it's done by the company called Texthelp  01:21:57.040 --> 01:22:02.960 and , you might be surprised but the the, the speech solution behind it  01:22:04.400 --> 01:22:11.200 is my system, the same as in MathJax so, so yes so that is good, yes.  01:22:13.360 --> 01:22:17.600 and it's, it's also very good for it has very good dyslexia support so they're  01:22:17.600 --> 01:22:20.640 their, their line of products aimed primarily at  01:22:20.640 --> 01:22:24.800 dyslexia support not, it's not so much at blind users.  01:22:26.960 --> 01:22:33.680 Okay, uh, we have two other questions. um one is about using an expanded UTF-8  01:22:33.680 --> 01:22:38.160 character set to represent math, especially for metadata shared within  01:22:38.160 --> 01:22:44.000 between systems. um what do you think of that? um  01:22:44.000 --> 01:22:47.120 so I assume you're meaning uh the, the upper  01:22:47.120 --> 01:22:52.160 plane or first plane and math alphabets in these things.  01:22:52.160 --> 01:22:57.840 um yes, there's nothing wrong with that. we can, and so in MathCheck we can use  01:22:57.840 --> 01:23:02.480 all of those. Um, in speech I can, they, they're all  01:23:02.480 --> 01:23:06.480 supported so there's, there's all the systems which  01:23:06.480 --> 01:23:10.240 do not go beyond uh the regular character sets but  01:23:10.240 --> 01:23:16.880 uh I support all of them. so it's all there .okay. yes yeah  01:23:16.880 --> 01:23:20.480 it's, it's a reasonable idea, yeah. you just have to be  01:23:20.480 --> 01:23:23.920 aware that not all of them are implemented everywhere  01:23:23.920 --> 01:23:26.240 so sometimes you get these little boxes which  01:23:27.680 --> 01:23:33.440 which then represents some, some uh some UTF-8 character on the on the upper  01:23:33.440 --> 01:23:38.240 unicode plane. yeah uh and then one, one other question  01:23:38.240 --> 01:23:42.480 and then I want to ask a kind of, a little bit higher level  01:23:42.480 --> 01:23:45.440 question, but um you showed the  01:23:46.800 --> 01:23:53.040 SVG files, and um this question was um, are all of the descriptions of the  01:23:53.040 --> 01:23:56.160 image examples embedded in the SVG file?  01:23:57.360 --> 01:24:03.200 yes they are. yep okay. so um think about, so the nice thing about  01:24:03.200 --> 01:24:06.720 SVG is you can do multiple things with it. you  01:24:06.720 --> 01:24:12.080 can have it very hierarchical, meaning that you can group semantically  01:24:13.040 --> 01:24:17.520 relevant elements together. so say for the chemistry, for instance, double bonds  01:24:17.520 --> 01:24:20.240 would be in a single, would the two lines would be in a group, and  01:24:20.240 --> 01:24:24.560 then the ring would be in a group, and so on and so forth, and you can embed,  01:24:25.760 --> 01:24:31.600 uh XML information in alternative namespaces and on that basis effectively  01:24:31.600 --> 01:24:35.680 you embed a graph structure as a shadow DOM, and  01:24:35.680 --> 01:24:37.520 that's all in there and then you need a bit of  01:24:37.520 --> 01:24:40.160 JavaScript magic to actually animate it.  01:24:42.720 --> 01:24:50.080 great okay. um so looking at the next, I don't know five,  01:24:51.440 --> 01:24:56.160 three to five years, um what advances do you expect  01:24:56.160 --> 01:25:00.720 to, or do you hope will happen with regard to STEM and math accessibility  01:25:00.720 --> 01:25:03.680 that, that we should look forward to. we've  01:25:03.680 --> 01:25:07.600 talked a lot about what we can do right now. what kinds of  01:25:07.600 --> 01:25:12.720 things do you think we'll be able to do in the future? Um. Um,  01:25:12.720 --> 01:25:16.320 well with so, so, so one thing I hopefully, you should be,  01:25:16.320 --> 01:25:19.040 one should be able to do quite quickly is, um  01:25:21.040 --> 01:25:23.600 generate tactile output more,  01:25:24.720 --> 01:25:29.040 more easily. I mean right now from what I understand it takes about six months for  01:25:29.040 --> 01:25:31.920 a textbook to be put into braille and Nemeth braille.  01:25:33.040 --> 01:25:36.400 so with the currentt workflow we're working on with, with pretext,  01:25:36.400 --> 01:25:40.960 um uh which, who do a lot of open source textbooks of course,  01:25:41.520 --> 01:25:48.880 um it should ideally, be the workflow. there might still be a need for a  01:25:48.880 --> 01:25:52.480 transcriber to manually correct some of the math and some of the  01:25:52.480 --> 01:25:56.480 the layout, but it should be a lot faster than setting everything by hand.  01:25:58.480 --> 01:26:01.280 so that's, that's kind of a short, short term goal.  01:26:01.840 --> 01:26:04.880 in terms of more long term, I would assume that, so  01:26:06.000 --> 01:26:09.280 for instance in MathCheck, we've now resurrected  01:26:09.280 --> 01:26:15.760 uh, somebody from Japan resurrected a, a graphing tool. It’s for the Xy-pic library.  01:26:18.560 --> 01:26:22.320 the next step for me will be to make that accessible.  01:26:22.320 --> 01:26:26.800 so obviously that's more of a research goal because I have no idea yet how  01:26:26.800 --> 01:26:30.880 one actually makes Xy-pic images and and things accessible.  01:26:30.880 --> 01:26:34.320 how do you explain a graph to somebody who can't see them, and these kind of  01:26:34.320 --> 01:26:37.280 things. TikZ hopefully will   01:26:37.280 --> 01:26:39.120 eventually be also on the agenda.  01:26:39.120 --> 01:26:42.800 um you can already produce SVG out of TikZ. It shouldn't  01:26:42.800 --> 01:26:48.240 really be a big issue, uh, to also add sufficient semantics to some of these  01:26:49.600 --> 01:26:55.200 expressions and, and get, get some of these things out. um  01:26:56.320 --> 01:27:00.320 yeah so, so from that point of view, I mean other, other things in other areas  01:27:00.960 --> 01:27:07.520 uh we've lost. or two years ago I start, organized a hackathon, um,  01:27:07.520 --> 01:27:11.040 at The Lighthouse in San Francisco where we started making a  01:27:11.040 --> 01:27:14.320 JupyterLab and Jupyter notebooks accessible.  01:27:14.320 --> 01:27:18.000 so, to have accessible computational devices out there in  01:27:18.720 --> 01:27:20.160 computational environments,  01:27:22.960 --> 01:27:27.520 in pat, in pretex they're sage computations which are partially accessible.  01:27:28.160 --> 01:27:33.600 some of you might be familiar with webworks. That, their accessibility is  01:27:33.600 --> 01:27:37.920 primarily based on the MathCheck accessibility. I understand they, they are  01:27:37.920 --> 01:27:40.400 quite happy with that. I have to admit that  01:27:40.400 --> 01:27:43.840 I know the people, but I haven't really used it very much but ANita you might,  01:27:44.560 --> 01:27:46.720 you might know them or, I don't know, somebody.  01:27:49.600 --> 01:27:54.720 So, hopefully the, the, the, I mean it will never be  01:27:54.720 --> 01:27:59.440 fully, fully solved, uh the math, the math question if you  01:27:59.440 --> 01:28:02.320 like, but uh it, it should improve uh  01:28:03.280 --> 01:28:06.880 and, and what I’m hoping for is that we get away  01:28:06.880 --> 01:28:11.840 from thinking k to 12 only and, and make more inroads into  01:28:11.840 --> 01:28:15.520 serious and more advanced mathematics, university,  01:28:16.160 --> 01:28:20.800 up to research mathematics where people can take the paper they get on archive,  01:28:20.800 --> 01:28:24.000 put it through that the processor, get tactile math  01:28:24.000 --> 01:28:28.640 out, where we can understand more of the semantics.  01:28:28.640 --> 01:28:33.120 um I've got currently got a research project,  01:28:33.120 --> 01:28:39.520 uh where we are trying to find out or mine context semantics from math papers  01:28:39.520 --> 01:28:42.640 and, and where you then try to understand, is  01:28:42.640 --> 01:28:46.400 something defined, can we use the definition in order to make that  01:28:46.400 --> 01:28:50.160 particular expression more, more accessible. say that this g  01:28:50.160 --> 01:28:55.840 is a group element rather than a functional one.  01:28:55.840 --> 01:29:03.760 Great. um so we're at time, and uh unless there are other questions,  01:29:03.760 --> 01:29:08.320 uh we really want to thank you for your time. I know it's late there in the UK.  01:29:08.320 --> 01:29:12.240 you also said you’re a night owl. thank you for spending  01:29:12.240 --> 01:29:19.200 um, some of your evening with us and, um for, for um talking with our audience  01:29:19.200 --> 01:29:21.840 about all different kinds of things. we have a lot of  01:29:21.840 --> 01:29:24.880 different interests, lots of different contexts, lots of  01:29:24.880 --> 01:29:28.000 interesting problems and different levels of knowledge, and  01:29:28.000 --> 01:29:31.440 we sincerely appreciate your, your time with us today.  01:29:32.240 --> 01:29:35.360 so thank you very much, and thank you also to  01:29:36.640 --> 01:29:40.560 the planners of the, the event and the panelists who  01:29:40.560 --> 01:29:42.720 came to bring moral support. so  01:29:43.600 --> 01:29:47.120 um yeah. well thank you very much for  01:29:47.120 --> 01:29:49.760 having me. I assume normally we would now start mingling  01:29:49.760 --> 01:29:53.600 with, for drinks, but uh I shall have a pint on my own then.  01:29:53.600 --> 01:29:59.440 Yes, enjoy a pint on your own um, and we will post the slides, on the link to  01:29:59.440 --> 01:30:01.760 your slides on, uh our slide   01:30:01.760 --> 01:30:09.520 link which is bitly slash Volker open ed. so  01:30:09.520 --> 01:30:11.760 we'll hopefully get that done, uh fairly  01:30:11.760 --> 01:30:14.800 quickly, so. They are there. all right thank you very much.  01:30:16.800 --> 01:30:20.000 all right thank you. bye-bye. thanks everybody, bye.