Experiments Towards Web 2.0 Accessibility

5 downloads 1148 Views 115KB Size Report
Sep 12, 2007 - ing more prominent within popular websites, such as Google and Yahoo ... relies on the latest browser technology being used and these.
Experiments Towards Web 2.0 Accessibility Elizabeth C Stringer, Yeliz Yesilada and Simon Harper Information Management Group School of Computer Science, University of Manchester Manchester UK

[email protected], [email protected], [email protected] ABSTRACT

Information within the Website is not always accessible but it is always available and this information, in some form, is transferred using HTTP. Therefore, we hypothesise that by monitoring network traffic, updates can be identified that are missed by screen readers. The goal of this experiment was to better understand how and when updates occur, how a screen reader copes with an update and how the network traffic reflects the update. We compared the output of the network analyser to the screen reader and found that more updates were registered with the network analyser and in some cases content was also identified. We conclude that our method could be used to inform screen readers when an update occurs and, where the content is present, give it instead of a refresh command leading to more information being available.

Client side dynamic updating within World Wide Web (Web) pages is becoming a prevalent feature across popular websites and services. The problem is that screen readers cannot identify when these updates occur. Our solution proposes that, by analysis of network communication, the underlying information can be extracted and presented to assistive technologies.

Categories and Subject Descriptors H.5.4 [Information Interfaces and Presentation]: Hypertext/Hypermedia—User Issues; H.5.4 [Information Interfaces and Presentation]: Hypertext/Hypermedia— Architecture

General Terms

2.

Human Factors

Keywords Web Accessibility, Web 2.0, AJAX, XForms, IFrames

1.

DYNAMIC UPDATING

AJAX [2], XForms [3] and IFrames[4] transfer all their content with a combination of JavaScript and HTTP. HTTP is a basic protocol that opens TCP/IP connections between the client and the server. It does not contain a unique identifier, instead, connection ports initiated by the client identify responses to their corresponding requests [5]. JavaScript works with most modern screen readers, however, the functionality varies and the update notification is poor. Even if an update is registered, the position of the update is unknown. JavaScript objects that require mouse interaction are not accessible to visually impaired people [6]. The Accessible Rich Internet Applications (ARIA) Roadmap aims to allow assistive technologies to interact with JavaScript powered Websites as a temporary solution until a declarative markup language can be developed [7]. There are several approaches made to make scripts more accessible and the part that governs updates is done by extending XHTML to include semantic information within the markup so authors can identify what operations might occur within particular sections of the page [7]. In ARIA, updating micro-contents are identified as regions with properties. These changes are currently being implemented in Firefox 3 [1]. This approach relies on the latest browser technology being used and these techniques being adopted by Web developers. Our project aims to instead provide a method that works with existing Web pages, browsers and assistive technologies which would complement the work done in ARIA. MutationEvents are Events triggered by a User Agent (UA) when the Document Object Model (DOM) changes. These should fire when updates occur to inform Assistive Technologies (AT) of updates [8]. However, even though ATs

INTRODUCTION

We define dynamic updating as updates to a Web page that occur when a user is operating the interface without appearing to change the Uniform Resource Locator (URL). These updating features can be described as independently updating micro-contents because they apply self contained changes to the Web page. These micro-contents are becoming more prominent within popular websites, such as Google and Yahoo, because they enable ‘Push’ updates and reduce network downloads compared to traditional Web pages. This is achieved through asynchronous updates that refresh parts of the page as opposed to the whole. There are a number of technologies used to achieve these updates: Asynchronous JavaScript and XML (AJAX), XForms and IFrames. However, accessibility issues arise with visually impaired users that rely on screen readers to access these websites [1]. As the content performs an update, screen readers fail to register this update and so cannot inform the user of the change [1]. By comparing the technologies that power these updates, it was determined that they have two aspects in common; they use JavaScript and Hypertext Transfer Protocol (HTTP) Requests. Copyright is held by the author/owner(s). HT’07, September 10–12, 2007, Manchester, United Kingdom. ACM 978-1-59593-820-6/07/0009.

33

pick these up, they have difficulty interpreting their meaning and context [9].

and combinations deal with dynamic updates and incorporate these findings into our approach.

3.

5.

EXPERIMENT

This paper presents an experiment that compares screen reader output to information obtained by a network analyser. We examined the output to see when updates occur, whether they are registered by the network analyser and/or the screen reader, and whether the content of the update was available. We used Wireshark1 as our network analyser and Guide2 as our screen reader. We used Yahoo Mail Beta3 because it is a Website that has good examples of dynamic updating. A parser was also created to process the output of the network analyser.4 . Packets, identified as HTTP by the parser, were analysed to see how the information contained within the packets corresponded to actions made when a specific area updated. These were compared with the ability of the screen reader to identify the updates and relay the content to the user.

4.

CONCLUSIONS

Our hypothesis is that by monitoring network traffic, updates can be identified that are missed by screen readers. Our keyhole experiment showed that screen readers have difficulties identifying updating content. By using a packet analyser, the majority of updates were identified that were otherwise missed. It also showed that in fifty percent of successfully identified updates, the content of the update was also retrieved. Of those that were not retrieved, the three scenarios that caused this were identified: caching, client side processing and security intervention. Using this approach, screen readers could be prompted to re-read the page. A larger scale experiment will now be conducted to ascertain what information is typical for different design patterns and where the issues mentioned cause most effect. In addition, other technologies such as XForms and IFrames and other screen readers such as JAWS and WindowEyes will be tested. Another study will be conducted with users to see which information is relevant to the user when interacting with Web 2.0 websites to judge the impact of these limitations.

DISCUSSION

From our experiments, it was observed that the network analyser consistently outperforms the screen reader for detecting updates to the page. When the screen reader did identify the updates, they were often inserted at the end of the page. During the reading, the screen reader identified items that were not visible on the screen. As hidden objects are not available for action until they are activated, this could be misleading to a user. This is common to all screen readers, not just the one used for comparison [1]. This experiment was performed to provide greater understanding of updating content. Our approach could be used to detect updates for the screen reader, and where content is available, present it instead of a refresh so that information is conveyed faster. The final application of this understanding is to aid in the creation of a black box that analyses network traffic and presents the information contained within it to screen readers in a similar way to ARIA. Unlike ARIA, however, the black box will not rely on any particular combinations of technologies. More analysis is needed to find a suitable way of presenting this information. There are also limitations to this approach in its present form. Currently, client side JavaScript operations cannot be identified. If the operation that triggers the client side script can be identified from the source code, there is a greater chance of being able to process the update by monitoring interaction with the page in addition to the network traffic. If the page has been previously cached, the initial state of the website cannot be identified. Therefore, we need to request the page again from the source website in order to present the full information. In the case of security measures, an examination of how browsers process these requests is needed to see if there is an independent way to access the information without breaking security protocols. For a truly independent system, we need to test the remaining technologies and alternative screen readers and browser configurations to determine how different implementations

6.

ACKNOWLEDGEMENT

This paper is written as part of the EPSRC funded SASWAT Project (EP/E062954/1).

7.

REFERENCES

[1] Gibson, B. In W4A: Proceedings of the 2007 international cross-disciplinary workshop on Web accessibility (W4A), (2007). [2] Mahemoff, M. Ajax Design Patterns. O’Reilly Media Inc, June (2006). [3] Boyer, J. M., Landwehr, D., Merrick, R., Raman, T. V., Dubinko, M., and Leigh L. Klotz, J. W3C recommendation, W3C, March (2006). http://www.w3.org/TR/xforms/ - Last Accessed 03/05/07. [4] Raggett, D., Hors, A. L., and Jacobs, I. W3C Recommendation, W3C, December (1999). http://www.w3.org/TR/html401/ - Last Accessed 03/05/2007. [5] Orebaugh, A. Ethereal Packet Sniffer. Syngress Publishing Ltd, (2004). [6] Thatcher, J., Burks, M. R., Heilmann, C., Henry, S. L., Kirkpatrick, A., Lauke, P. H., Lawson, B., Regan, B., Rutter, R., Urban, M., and Waddell, C. D. Web Accessibility Web Standards and Regulatory Compliance. Springer-Velag, (2006). [7] Schwerdtfeger, R. W3C Working Draft, W3C, December (2006). www.w3.org/TR/aria-roadmap/ Last accessed 03/05/2007. [8] Pixley, T. W3C Recommendation, W3C, November (2000). http://www.w3.org/TR/DOM-Level-2-Events Last Accessed 13/06/07. [9] Thiessen, P. and Chen, C. In Proceedings of the 2007 international cross-disciplinary conference on Web accessibility (W4A) W4A ’07, 7–14. ACM Press, (2007).

1

Wireshark - http://www.wireshark.org Guide - http://www.softwareexpress.co.uk/ 3 Yahoo Mail Beta - http://mail.yahoo.com 4 More details and results available at http://saswat. cs.manchester.ac.uk/ 2

34