Upload
markel-vigo
View
6.207
Download
4
Embed Size (px)
DESCRIPTION
Citation preview
Evaluating Web Evaluating Web Accessibility for Specific Accessibility for Specific Mobile DevicesMobile Devices
Markel Vigo, A. Aizpurua, M. Arrue and J. Abascal
Laboratory of HCI for Special Needs
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
University of the Basque Country
IntroductionIntroduction
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Generally, web content is developed with desktop computers in mind Server-side services or proxies transform content User Agents provide new features for better user
experience: fast navigation mechanisms or content linearization
• Mobile Web and Web Accessibility for physically impaired users share similar problems
• Problems that able-bodied user may have are similar to those found by people with disabilities
Guidelines Architecture Case Study ConclusionsIntroduction
Mobile Web vs Web AccessibilityMobile Web vs Web Accessibility
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
Mobile Web Web accessibility
Small display size causes disorientation on the user
Lack of context disorientates visually impaired users
Lack of pointing device forces users to use keyboards. Navigation is slowed down
Screen reader users suffer information overload: navigation bars and menus.
Typing is a tedious task due to low text input rate
Users with motor disabilities face analogous problems
Due to low bandwidth images tend to not to be loaded
Not providing alternatives for visual content raises accessibility barriers
Lack of colour support may cause information loss
Information conveyed with colour causes problems to colour-blind users
Lack of support causes information loss
Assistive technologies tend to have problems with newer technologies
Guidelines Architecture Case Study ConclusionsIntroduction
GuidelinesGuidelines
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Mobile Web Best Practices MWBP 1.0 were released by the W3C
• mobileOK test are techniques to conform with MWBP 1.0 mobileOK Basic: mobileOK Basic conformance “functional
user experience” mobileOK Pro: techniques are to be released
• Rely on the “Default Delivery Context”
Usable screen width: 120px minimumMark-up language support: XHTML Basic 1.1Character-encoding: UTF-8256 colours minimumCSS level 1 supportScripting is not supported
Guidelines Architecture Case Study ConclusionsIntroduction
Guidelines & EvaluationGuidelines & Evaluation
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Related work: there are several tools that evaluate web pages against mobileOK Basic tests
Guidelines Architecture Case Study ConclusionsIntroduction
TAWTAW MWIMWI
ready.mobiready.mobiEvalAccessMOBILEEvalAccessMOBILE
Guidelines & EvaluationGuidelines & Evaluation
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• mobileOK Basic test are intended for development purposes BUT
• There are different flavours of mobile devices
• These test produce problems when mobile devices deviate from DDC Newer models false positives Legacy devices false negatives
Guidelines Architecture Case Study ConclusionsIntroduction
Guidelines & EvaluationGuidelines & Evaluation
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• MWBP 1.0 are full of statements regarding device features. E.g.: “Do not use tables unless the device is known to support them”
• Objective: a tool that evaluates mobileOK Basic tests considering the specific features of mobile devices when required
• mobileOK Basic test are extended, focusing on the device-dependent tests
Guidelines Architecture Case Study ConclusionsIntroduction
Architecture: overviewArchitecture: overview
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Source Code Retriever
• Device Information Retriever
• Evaluation Engine
Device-Tailored
Evaluation
Device-tailoredReport
Guidelines Architecture Case Study ConclusionsIntroduction
Mobile Device’s brand name and
model
Source Code RetrieverSource Code Retriever
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Objective: retrieve the source code in the same way that a mobile device would get it
• Some web servers identify the ‘user-agent’ HTTP header and deliver different web content
Source Code Retriever
XHTML file
WWW
• The Source Code Retriever retrieves a web resource simulating the access of a determined device by manipulating the “user-agent” HTTP header
Guidelines Architecture Case Study ConclusionsIntroduction
Device Information RetrieverDevice Information Retriever
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Two ways to obtain information about a determined mobile device: UAProf profiles extended CC/PP profiles using prf
namespace WURFL (Wireless Universal Resource file) XML file
• Heterogeneous information sources
CC/PP file
Device Information
Retriever
Jena
WURLF API WURLFprofiles
UAPRofprofiles
• Required data are extracted and a CC/PP file is created
• Both sources have complementary information
Guidelines Architecture Case Study ConclusionsIntroduction
Extending mobileOK Basic testsExtending mobileOK Basic tests
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• The implications that devices characteristics have on the mobileOK Basic tests have analyzed
• New CC/PP based vocabulary has been created to express certain concepts
• Examples CONTENT_FORMAT_SUPPORT access:picFormatSupport NO_FRAMES prf:FramesCapable OBJECTS_OR_SCRIPTSprf:JavaScriptEnabled
• Information to be retrieved are the issues that the DDC captures: character encoding, image format support etc.
Guidelines Architecture Case Study ConclusionsIntroduction
Guidelines InstantiatorGuidelines Instantiator
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• mobileOK Basic test are implemented using UGL (Uniform Guidelines Language) It has slots so that values can be put in guidelines
• Device data from the CC/PP file is used to fill in slots in the mobileOK test
• Once the guideline is completed XQUERY tests are dynamically created
XQUERY tests
mobileOKtests
CC/PP file
Guidelines Instantiator
Guidelines Architecture Case Study ConclusionsIntroduction
Guidelines InstantiatorGuidelines Instantiator
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Example for the IMAGE_MAPS test
XQUERY tests
mobileOKtests
CC/PP file
Guidelines Instantiator
<access:pntSupport>true</access:pntSupport>
<test_case id="8"><evaluation_type>auto</evaluation_type><evaluation_result>error</evaluation_result><profile_feature type="access:pntSupport"/><value> </value><element><label>OBJECT</label><test_elem>check attribute</test_elem><related_attribute><atb>ismap</atb></related_attribute></element></test_case>
CC/PP excerptCC/PP excerpt
UGL excerptUGL excerpt
let $tmp:=web_doc.xml//OBJECT[@ismap] return if(not( ))thenfor $i in $tmp return<error>{$i/@line, $i/name()}</error>
XQUERY testXQUERY test
true
true
Guidelines Architecture Case Study ConclusionsIntroduction
Evaluation EngineEvaluation Engine
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Summary:1. Web resource is retrieved changing HTTP
headers2. Information regarding the device is
retrieved from profile repositories and a CC/PP file is created
3. Slots in guidelines specifications are filled in with CC/PP data and XQUERY tests are automatically created
4. XHTML file is evaluated against the dynamically created queries and a device-tailored report is obtained
Evaluation Engine
EvaluationReport
Source Code Retriever
XHTML file
WWW
XQUERY tests
mobileOKtests
CC/PP file
Device Information
Retriever
Guidelines Instantiator
JenaWURLF
APIWURFLprofiles
UAPRofprofiles
• Once we have a set of XQUERY tests evaluation is straightforward using XSLT processors
Guidelines Architecture Case Study ConclusionsIntroduction
Case StudyCase Study
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Nine web pages have been evaluated with 3 different mobile devices D1: less support than the DDC D2: similar to the DDC D3: more features than the DDC
mobileOK Basic Specific Evaluations for mobileOK Basic
D1, D2, D3 D1 D2 D3
www.google.com 4 4 4 4
www.youtube.com 3 2 3 2
www.flickr.com 4 4 5 4
www.amazon.com 9 10 9 9
www.gmail.com 9 5 9 5
www.facebook.com 7 7 6 6
m.yahoo.com 6 6 6 6
m.twitter.com 8 6 8 6
www.wikipedia.org 428 363 358 280
Guidelines Architecture Case Study ConclusionsIntroduction
Case StudyCase Study
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Looking carefully at results… Mobile versions of traditional web pages have fewer errors with
respect to desktop web pages
Devices with less support than the DDC tend to produce more errors false negatives ↓
while those with better support and more characteristics yield fewer errors false positives ↓
Guidelines Architecture Case Study ConclusionsIntroduction
Last remarksLast remarks
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• The tool can be used for the development of accessible applications
• Main purpose is to plug this tool in a more general framework Goal: obtain user and device tailored accessibility scores as
the user interacts
• Demonstration of a prototype: http://sipt07.si.ehu.es/mobile/
Guidelines Architecture Case Study ConclusionsIntroduction
Conclusions & Future WorkConclusions & Future Work
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
• Summary: The presented tool deals with device specific accessibility
issues mobileOK Basic tests have been extended False positives and false negatives diminish
• Future Work mobileOK Basic tests contain many references to HTTP
headers content. Currently our approach deals with mark-up issues
mobileOK Pro are expected to be released
Guidelines Architecture Case Study ConclusionsIntroduction
Evaluating Web Evaluating Web Accessibility for Specific Accessibility for Specific Mobile DevicesMobile Devices
Markel Vigo, A. Aizpurua, M. Arrue and J. Abascal
Laboratory of HCI for Special Needs
International Cross-Disciplinary Conference on Web Accessibility, W4A 2008
University of the Basque Country
Any question?