13:50:36 RRSAgent has joined #dc2005 13:50:36 logging to http://www.w3.org/2005/09/14-dc2005-irc 13:50:48 rrsagent, make logs world 13:51:19 ... 13:51:36 Thailand is now requirting accessibility in government pages 13:53:16 ... 13:54:29 (scribe is busy trying to explain where we are, and people are introducing themselves 13:55:33 will try to catch up in a bit... 13:59:07 achuter has joined #dc2005 13:59:25 liddy: we have changed the agenda around.... 13:59:50 ... we have Shadi from W3C to talk about EARL, Vicente Luque talking about automated testing for the guidelines 14:00:09 ... looking at how we can make them work as much as possible easily for people. 14:00:32 ... We are looking to make sure that WAI work, IMS work, DC work, and other accessibility work is actually compatible. 14:01:06 ... We are not saying that these things are not the best, but we are also recognising that there is a very big world out there that we are trying to deal with, and we need to keep checking that. 14:02:04 Seems that pretty much everyone has an understanding of what accessibility is about, at least. 14:02:29 Michael: at ETSI we are bundling accessibility and children together because they have similar issues 14:02:54 liddy: (scribe's grandmother!) don't forget the grandmothers as a classic example 14:05:10 Martin: giving an overview of the IMS accessibiltiy for all stuff. Starts by teasing Australians about the cricket :-( 14:05:59 ... basic idea is to have interoperability between different accessibility standards. 14:06:37 ... biggest problem is diversity of needs and preferences, so two people who are the same will each be more efficient working in a different way 14:08:14 ... W3C WCAG approach aims to have an ideal situation - making content universally accessible. This approach looks at how to augment things taht start out not being accessible, perhaps even by substituting a different resource. 14:08:59 ... for example no point teaching a blind person about visual perspective by showing them pictures, so it might make as much sense to substitute with some specialised resource. 14:09:27 ... at open University we have about 2-3 thousand students with disabilities enrolling each year. 14:09:47 ... so how do we find out if the courseware for each of them is accessible to each of them. 14:10:37 scribe: chaals 14:11:15 martin: want the system to understand each user's needs, and where relevant be able to maintain the information. 14:11:24 ... metadata is key to this approach. 14:12:20 ... two types of accessibility metadata: "this content is WCAG double-A conformant", or 14:12:46 ... user profiles, to allow matching between what the user wants/needs and the content they are looking for. 14:13:46 chaals has joined #dc2005 14:14:02 Q: where does the data about users live? 14:14:29 A: No requirement for it to be identifiable, and it is able to be kept in various different ways. 14:14:57 ... the privacy issue has been considered and is recognised as important. The implementations out there show various possibilitiers 14:17:41 martin: this allows for the real world. Some people produce content, and an ancillary service makes an accessible version. 14:18:35 ... as liddy said, no conflict between objectives of AccessForAll and WCAG 14:18:55 ... I don't entirely believe in universal design. 14:20:02 ... I like the idea, but it doesn't always work out in practice. AccessForAll is about how to achieve the WCAG goals (and meet their requirements) in a different way 14:20:36 [chaals: effectively AccessForAll describes a lot of detail about how to meet WCAG through meeting checkpoint 11.4, based on a starting point of non-conformant content] 14:21:12 martin: two parts. AccLIP is the user profile, and AccMD is the document profile. 14:21:19 lnevile has joined #dc2005 14:21:56 ... started out including XML bindings, but there are many different ways of including the information. 14:23:38 I have uploaded the slides but cannot find them!!! 14:24:16 chair: liddy 14:25:41 martin: you can define links between resources, so if you find the captions their metadata points to the video and vice versa 14:26:30 ... you can define transformation information - changing the presentation of teh content to match the user profile 14:26:57 ... or you can describe how to tailor the interface the user has, to meet their needs 14:27:14 Q: What if users don't think of themselves as having a disability? 14:27:37 A: These are just resources in a learning object collection. Searching for something brings up a version of it. 14:27:55 Q: Are these transformations done on the fly, or in advance 14:28:15 A: It's up to the implementation. You don't have to do it in advance, but you can. 14:28:41 liddy: the value of it is that when you didn't anticipate anything, you can find stuff that is used to help transform. 14:29:21 chaals: So this is potentially completely distributed - you don't have to collect all these things yourself? 14:29:27 A: right 14:30:04 Q: What about Math, etc? 14:30:31 A: It is a difficult problem - not how to describe the problem, but what you should actually end up with to make things work 14:30:55 martin: in the UK, legislation is anticipatory - you have to anticipate that you will have pople with all kinds of disab ilities 14:31:20 More than distirbuted, it allows for creation of accessible components on the fly or even 'within 24 hours' ... 14:31:22 helle: In Denmark we have problems with copyright blocking this. You can provide stuff to blind students but you cannot share it 14:31:55 martin: We don't register who is blind, so it is a question of convincing a lawyer in a given case that this system is giving stuff to blind people 14:32:47 mary: in the US students are required to buy the book to get access to an accessible version. Publishers will only allow registrationof requests for electronic or alternative version. Each time a student needs soething, they have to request again - it isn't automatic. 14:33:03 martin: That is better contained in fact in other metadata that handles legal information. 14:33:17 Martyn's presentation is available at http://www.ozewai.org/dc2005-cooper.ppt 14:33:37 no - ftp is blocked from here! sorry 14:34:00 s/is available/will be available/ 14:35:01 martin: this could be pulled together in several ways. W3C device independence? or other? 14:35:39 ... there are a lot of groups working on this, but there is a lot of overlap in the actual people involved in these groups. 14:36:00 ... working to move from specifications to international standards. 14:37:21 ... drafts of ISO work available via http://dublincore.org/accessibilitywiki 14:38:03 liddy: these will be updated next week or so, following from this discussion. They are currently the most up-to-date. 14:38:27 ... if you have comments you can go through ISO, but you can also send them informally and we do actually look at them. 14:38:53 michael: There are other groups working on accessibility metadata, but don't always realise it. 14:40:01 ... but it is also nice to adoopt and comment instead of re-invent - that is the approach we have taken at ETSI 14:40:59 martin: this is all being done in english, but we need to make this work in a multilingual world... 14:41:28 ... final comments: accessforall seems like a powerful appraoch, but ongoing work for wide adption is necessary 14:42:44 test 14:43:05 shadi: shadi will begin presentation on EARL 14:44:27 works at the W3C, involved with the WAI guidelines. 14:44:48 guidelines, methods, conducting outread and education all for accessibility initiatives 14:44:59 today: evaluating web site accessibility 14:45:13 there is no formal grammar for non-text content. 14:45:27 e.g. text equivalnets for non-text content 14:45:33 lables and descs for content 14:45:38 layout structure and reading order 14:45:46 color contrast and visual appearance. 14:46:33 however, it i s more complicated: what is a text equivalent for an image, i.e. difficutly with logos vs. other images. 14:47:15 web developer requirements: designers, content authors, rogrammers, managers 14:47:39 all have different responsibilities and approaches to web accessibilities. 14:48:15 thus, diverse types of user interfaces, diverse functionality and usages, diverse web developer reqs, diverse capabilities and quality. 14:48:38 i.e. tools are very different in the quality they provide which makes things difficult. 14:48:51 EARL should be a machine-readable syntax 14:49:19 for text results. in this way you can provide rich metadata about test results. 14:49:45 also, provide rich metadata about test results. 14:50:04 royalty free, platform independant and vendor neutral. 14:50:14 purpose is generic quality assurance 14:50:38 but, very important, to achieve web accessibility (on toop of all this) 14:50:49 EARL is an RDF schema with Classes. 14:50:58 Assertor: carries out the evaluation 14:51:03 Subject: object being evaluated 14:51:15 TestCase: test criteria for evaluations 14:51:30 Result: pass / faill / Not Available, message, etc. 14:51:39 <---- basic structure of EARL 14:52:05 also there is "Software" which provides the test results. 14:53:04 these classes are comparable with the core of the Dublin Core... however, there is only 4 :0 14:53:06 :) 14:53:25 Benefits: being part of the larger RDF community 14:53:51 what you get (for free): reuse widely accepted vocabs. 14:54:03 reuse parsers, reasoners, interfaces. 14:54:12 query results (also on the metadata) 14:54:25 allow flexible vocabulary extension. 14:55:04 uses dublin core, FOF (friend of a friend) 14:55:10 what is RDF basically... 14:55:16 Resource Description Framework. 14:55:35 Charles draws a diagram explaining RDF. 14:55:59 Charles: say a thing has a relation to another thing to another thing, etc. 14:56:17 in a machine readable way. 14:56:39 back to shadi: a machine readable language 14:57:27 in EARL, in RDF we have an Assertor, Query, Results, and extend the vocabulary of the results and the relations 14:57:50 vocabular can be exteneded, e.g. " nearly passed" 14:57:56 Use cases of EARL: 14:58:02 collect test results from different tools 14:58:10 analyze, compare, prioritize assertions 14:58:32 banchmark tools to improve performance (makes developers nervous :)) 14:58:39 generate reports from aggregated results 14:58:47 integrate evaluation into developemnet tools 14:59:00 annotate and describe content conformance 14:59:16 last point is most relevant to DC 14:59:20 relevance for DC 14:59:32 annotate and describe conformance of resources 14:59:42 e.g. is the resource keyboard accessibile. 14:59:51 but also, who carried out the test, how reliable are they? 15:00:07 which tools have been used for testing? 15:00:20 soon: has the content chnaged since some date? 15:00:28 more work to be done by the WG in W3C 15:00:40 Status of EARL: still a working draft. 15:00:54 9 sept. 2005 EARL 1.0 Schema draft! 15:01:09 to come soon: EARL 1.0 Guide and requirements. 15:01:16 Open questions to resolve: 15:01:24 describing location for test results. 15:01:37 improving persistence of test reports 15:02:01 currently, only compared by date. 15:02:23 need something else which is less expensive, i.e. don't want to throw away results. 15:02:37 confidence levels for claimed results. 15:02:56 notion of confidence levels still controversial 15:03:12 there is a way in EARL to give a result, but say you have no idea what you are talking about :) 15:03:18 Get involved with EARL: 15:03:27 read and comment on ERT WG 15:03:39 join and contribute to the ERT WG mailing list. 15:03:50 everyone is welcome 15:04:07 of all types of participations. 15:04:15 thank you. this is the end of the presentation. 15:04:32 break o'clock. back in hallf an hour 15:32:38 we are back - 15:32:54 Vicente will be presenting his work and his slides are at http://www.it.uc3m.es/vlc/www2005poster.pdf 15:34:45 chaals has joined #dc2005 15:35:14 http://www.it.uc3m.es/vlc/www2005poster.pdf 15:35:29 liddy: questions and comments from before the break? 15:35:36 Let's have some quesitons and comments 15:35:54 scribe: lnevile 15:36:03 Mike: we have seen a lot of person to resource matching here - what about person to person matching - can we use the same approach? 15:36:53 chaals has changed the topic to: Dublin Core 2005 accessibility session until 19h00 Madrid time 15:37:07 David: the 'r4esource' in our model could probably be a person 15:38:11 Stephen: what about people using symbols to communicate? it's their language not a preference.. 15:40:01 Helle: with this kind of adaptability approach, I am a little concerned that we still need basic requirement that the sit4e contains what will be accessible... 15:42:56 Shadi: this seems to suggest that the owner of the website does not ened to worry any more 15:43:15 Charles: I don't see the distinction... 15:45:01 Liddy: suppose a set of universities get together and share resources . 15:45:49 Shadi: but is your resource conformant if you don't know that I provided captions - the user gets them but should you get credit? 15:46:48 David: Martyn suggested that the universal accessibility concept does not guarantee access for all - so the metadata helps. 15:47:33 Shadi: the ref. to the DRC report confuses people 15:48:55 Liddy: we separate the accessibility of the resource and the service - metadata helps make the service accessible 15:49:47 Charles: often people try t make their work look good by making it look different - the real difference bet A4A and WCAG is that A4A gives a good description of the resource about which you may claim accessibility status 15:50:47 Charles: when real content is not right, you can use A4A to build a description of something that will be accessible - independent of claims 15:51:25 Charles: A4A focuses on the individual's needs 15:52:28 Charles: WCAG does not help decide if the resource suits you - it provides a general accessibility claim 15:53:02 Helle: we have tried to avoid people making text-only versions. By promoting this idea, are we encouraging webmasters to be more slack. 15:55:14 David: is it the 'alternativeTo' that causes problems ...A4A gives you a metadata description of the accessibility properties of the resource ...so you can see what you've got 15:55:27 Liddy: that's like eating in Spain 15:55:44 Shadi: that's what WCAG already says... 15:56:18 Shadi: many want more built into HTML etc so this can be part of what is done.. 15:58:07 Shadi: it seems to take away the responsibility of the owner to provide signs 15:59:51 David: is it an issue that the responsibility is on more than one place? 16:00:22 Charles: it is the political message stuff that we are worrying about... 16:02:56 Charles: we can go round in circles without making progress 16:05:12 Mary: We have resistance from teachers, who are busy trying to teach and see the requirements as taking away from their ability to do that, and interfering with their academic freedom. 16:05:41 liddy has joined #dc2005 16:05:46 Liddy: There are things you can do with TILE, swapping between setups, that you can't do any other way 16:06:04 [chaals actually disagrees, but doesn't see the point in arguing about it] 16:06:26 Martyn: We have had some success using template-based approaches. 16:07:22 ... this gets people a long way there, with tools that automatically do a lot of the work. 16:08:12 Mary: most departments are hiring students or other cheap labour to do stuff at the lowest cost possible 16:08:23 liddy: and lots of people hate templates 16:08:44 Helle: TILE looks interesting, but you don't have the issue of having to go across universities 16:09:01 Liddy: Right, but it is meant to go across universities. 16:09:35 ah. Now we actually get to hear Vicente... 16:11:02 The slides are at http://www.it.uc3m.es/vlc/www2005poster.pdf 16:14:02 Vicente: basic problem is that there are a range of testing tools, using different tests, so they give different results 16:14:38 ... it would be nice to have formalised tests that use standards, which can provide very cheap testing. 16:19:02 ... (Tests a couple of pages in a few tools, and finds that according to Hera there are a couple of failures in http://www.w3.org/WAI for double-A and triple-A WCAG conformance 16:19:37 ... to do with CSS, validity, and some "until user agents" checkpoints) 16:20:12 ... Shows some xpath / xquery expressions that can be used for testing. 16:30:17 Liddy: next thing to coonsider is the charter that has been on the wiki for some time 16:30:41 chaals: Have you looked at describing what your rules actually test, so that it becomes possible to collect up different rules in a tool, and build the "best tool available" on the fly. 16:31:07 Mary: Is there a move to develop more semi-automation or automating more things in evaluation tools? 16:31:15 chaals: yes 16:31:39 shadi: there is a lot of work, but some things probably won't be automatable. 16:33:46 liddy: Moving to the charter of the group 16:34:21 ... propose a task group which is open, for anyone who is interested in the detailed day-to-day work, reporting to the larger group on what is new. 16:35:02 ... producing examples, best practice notes, and finalising the model and terminology, develop controlled vocabularies 16:35:17 ... seek formal liaison with ISO/JTC ... 16:35:39 ... develop relevant tools and profiles 16:35:48 Approved. 16:36:32 Liddy: there are weekly meetings Wednesday, 4pm US East Coast, 10pm central europe, 6am Thursday in Australia. 16:36:53 ... work is on the wiki always - http://dublincore.org/accessibilitywiki 16:37:15 ... there is also stuff going on in ISO, and it is actually possible to participate. 16:38:03 Mary: Didn't get the notifications 16:38:35 Liddy: Yeah, you need to register for it or it doesn't bother letting you know that you won't get mail. 16:41:12 chaals has changed the topic to: done... 16:41:22 rrsagent, make minutes 16:41:22 I have made the request to generate http://www.w3.org/2005/09/14-dc2005-minutes.html chaals