OSS-Based Development of a Web Accessibility Observatory

Recent Updates:

New Scientific Paper:: Towards a score function for WCAG 2.0 benchmarking 2012
New Blog Post:: United Nations Global E-government survey 2012 2012-03-01
Ph.D. Thesis:: Towards Automated eGovernment Monitoring

General Information

Download OSS-Based Development of a Web Accessibility Observatory as PDF (51 KB) .

Title: OSS-Based Development of a Web Accessibility Observatory.
Author(s): Terje Gjøsæter, Morten Goodwin, Parastoo Mohagheghi.
Published date: June 2006.
Published at: Second Workshop on Web Accessibility and Metamodelling 2006

Abstract

The EU-funded project EIAO (European Internet Accessibility Observatory) is establishing the technical foundations for a large scale observatory for monitoring the accessibility of European web sites.We will in this paper focus on presenting a heterogeneous architecture implemented for the EIAO web accessibility observatory. The architecture presented in this paper has been made possible through the use and reuse of Open Source Software (OSS) components. While OSS operating systems and desk top applications are widely used, there have been few publications on benefits and challenges of reusing OSS components for application development. Due to the availability of code and the right to modify it, OSS components may be modified and extended to fit the application requirements. The quality is also raised by using the resources of the OSS community for bug fixing or reporting fixes back to the community. This abstract presents examples of components that are central to the Observatory application.The observatory's software architecture contains components developed from scratch and based on OSS components. The goal has been to reuse OSS as far as possible to shorten development time. Only when no suitable OSS component has been found, new components are developed from scratch. We may therefore divide the components into two main categories; new, and OSS-based components.

New Components These components are developed from scratch more or less without extending any external components. Note that these components use functionality from other components, but do not extend them. An example of such a component is the URL directory. Even though the URL directory uses already developed OSS components, the functionality in this component does not extend any of them.
OSS-based Components These components reuse, modify, customize or extend already developed OSS components. An example of reuse without modification is reusing MySQL, while an example with heavily modification of an OSS component is the Crawler component, which is largely based on the Open Source web crawler HarvestMan. The main OSS-based components of the architecture, with notes on how each component benefits from reusing existing OSS components, are briefly described below. All these components (Crawler, WAMs) consists of the mentioned reused OSS, and of some new subcomponents that are not mentioned here.
Crawler The crawler downloads web sites, sends the pages to the Web Accessibility Measurement Components (WAMs) for evaluation, and stores the results in the RDF repository. The source code of the crawler expanded the HarvestMan crawler and thus saving development time compared to implementing a new crawler.
WAMs (Web Accessibility Metrics) The Web Accessibility Metrics is a set of formal rules created for reporting barriers of web pages received. The WAMs currently use Relaxed as a framework and Schematron for the formal rules. Even though the Relaxed framework has been customised, most of the work lies in specification of the rules. Relying on this Open Source framework, the amount of work was limited to only extending Relaxed to fit our needs, such as returning EARL/RDF. EARL is a W3C developed evaluation and reporting language intended to facilitate the exchange of test results between Web accessibility evaluation tools.
RDF Repository The RDF repository is a heavily customised version of 3Store. During the integration of 3Store some limitations were discovered such as missing the possibility of deleting any content. In order to do this, we needed to investigate the functionality of 3Store. Due to the OSS license of 3Store, we were able to merge a lot of the 3Store functionality from C to Python without performance loss and add the missing functionality. The RDF repositories are used to store the EARL assessment reports, URL list of the sites to crawl and current state information. The RDF repository also utilises Redland to, among others, translate RDF/XML to RDF triples. During the implementation of the repository, a small but blocking error in the handling of Unicode in Redland was discovered. Because the source code was available, we could adjust the code in the matter of minutes which was given back to the Redland community. This could not have been possible with the limitation of closed source code.
Data Warehouse The data warehouse uses the OSS database PostgreSQL and is developed to give flexible analyses of the data and to be able to provide these analysis through a Graphical User Interface (GUI).
Graphical User Interface (GUI) The Graphical User Interface will present the measurements locate in the data warehouse. We plan to use the OSS Content management System Plone written on Zope as a framework for this interface. Because Plone/Zope is Open Source and easily customisable, we can add adjust the Plone interface to e.g. be more accessible than standard Plone.

As shown above, the software architecture has been made possible through the use and reuse of OSS components. This includes the new components, which use open source components without modification, and the OSS-based components, which customise existing OSS components. The OSS components offer a variety of functionality central for the development of the observatory. One main criteria for OSS selection is the readability and understandability of code for modification and the quality of documentation and support. Thus, we have benefited from the availability and openness of Open Source software in multiple ways.For the next release of the EIAO Observatory, we plan to do a more thorough test of needed functionality of the integrated third party OSS components. We hope this will limit the number of unknown bugs and improve the quality of the software, while the architecture is mainly stabilized by now.

BIBTEX

@article{wwam2006ossdev,
    author{Terje Gjøsæter, Morten Goodwin Olsen and Parastoo Mohagheghi},
    title{OSS-Based Development of a Web Accessibility Observatory},
    booktitle{WWAM},
    year{2006},
    month{June}
}

The author of this document is:
Morten Goodwin
E-mail address is:
morten.goodwin ASCII 64 tingtun.no
Phone is:
+47 95 24 86 79