Site Navigation
Contents of Thesis ack'ments - Preamble - Introduction -Accessibility - W3C/WAI - LitReview - Metadata - Accessibility Metadata - PNP - DRD - Matching - UI profiles - Interoperability - Framework - Implementation - Conclusion - References - Appendix 1 - Appendix 2 - Appendix 3 - Appendix 4 - Appendix 5 - Appendix 6 - Appendix 7
Given the Dublin Core as a base for international, cross-domain metadata, it seemed obvious when the author started work with accessibility and metadata (in ???) that the two should be combined. First, it is necessary to establish that there is a reasonable chance that there will be metadata about accessibility of resources. It is also important to know if there are, as proposed, Web services that adapt resources for users. Finally, in this chapter, the relevant pre-history of the DC AfA accessibility metadata work is explained.
There is always, in the mind of metadata experts. the experience that shows that metadata is expensive to produce and that it is very often inaccurate. For this reason, it is important when proposing a new use or context for metadata, to be sure that it is necessary, not overly-complicated, and likely to be created and used. This section locates the current research in a world that is already partially prepared for it. Showing that there is a substantial amount of discoverable material in a range of formats suitable for people with varied needs and preferences, is important if there is to be more work in finding a way to describe the necessary needs and preferences and the resources that might satisfy them. Thus, the quantity of discoverable material is indicated within this section. In addition, unless the new descriptions can be used alongside those already in use, that is, unless there are existing descriptions that are interoperable with the new ones, there is not much point in undertaking the research. What follows shows that there is sufficient material and provides a base against which the new metadata might be tested for interoperability.
In the UK, the Royal National Institute for the Blind (RNIB) has developed and maintained the National Union Catalogue of Alternative Formats. Ann Chapman, in "Library services for visually impaired people: a manual of best practice" (2000), states that only 5% of the 100,000 new British titles published each year were converted into alternative formats. She points out that these formats were created by a range of individuals and organisations and made available in a number of different ways and places. "In 1989 R.N.I.B. began the process of computerising its card catalogues, thereby creating the National Union Catalogue of Alternative Formats (NUCAF)". Prior to this date, the RNIB had a catalogue of its own conversions and for five years prior to the establishment of the NUCAF, was spasmodically collecting catalogue records from others. She continues,
As part of the Department for Culture, Media and Sport funded programme to improve library and information services to visually impaired people, the role of NUCAF was reviewed in 1999 (Chapman, 1999). The review concluded that a national database of resources in alternative formats was an essential tool in service provision and that while NUCAF in its present form had limitations, particularly in respect of access, it did provide a good basis for a more comprehensive database of resources." ...
It further recommended that the new database should primarily cover the output and holdings of the specialist non-commercial sector, and that collaborative agreements with existing databases and union catalogues should be developed to cover the commercial sector publications." The review pointed out that "In addition to libraries, a range of agencies (doctors, dentists and health professionals, banks, advice centres, electricity, gas and water companies, tourist offices, schools and academic institutions, government departments, and service providers of various kinds) would either use the database or refer people to it. Currently visually impaired people and those working to support them are restricted to a few narrow avenues of access to NUCAF. The new database will be designed to be far more widely accessible to end users and library staff. To achieve this it was recommended that the national database should be held on a web-based system, supported by CD Rom and electronic file versions.
Eventually, as a result of various funding opportunities and projects carried out in a number of places, NUCAF was merged into a new service called REVEAL. In "Project One part A: The future role of NUCAF and a technical specification of the metadata requirements", Chapman (1999) reported "The national database should where possible use national and international standards. It should use the UKMARC format and conform to AACR2. Current RNIB subject indexing should be used for subject indexing, and LCSH entries retained where they exist in the records for the original items. A single set of headings for fiction genre/form should replace the existing ones. A full set of the data elements required has been identified."
These were found to be:
While NUCAF had catalogue records for many items, they were only items converted for the benefit of users with vision disabilities and they did not include representations in all formats or modes of access. Initially, they did not include commercially produced formats and they were expected to be catalogued only so they could be discovered, as was typical of the understanding of the use of metadata at the time (1999). The MARC21 007 fields provide for quite specific information about the form of tactile representation of information such as that it is contracted Literary Braille or 'spanner short form scoring' of music.
At 2.4, Chapman points out that the existing NUFAC's "only clearly defined objectives are those that relate to stock management and production management at the RNIB. It is therefore difficult for it to satisfactorily address functions outside the RNIB". She asserted that given the difficulties associated with copyright with respect to the transformation of information into alternative formats, the new data base would need to do more. She did not think of computers at that time as being able to automatically decompose information resources and recompose them to suit the needs and preferences of users. Her final recommendations included that, "The database must provide data rich bibliographic records".
At the time, the Library was UK’s most comprehensive collection of material on the subject of visual impairment. The resultant REVEALWEB, at the beginning of 2006, boasts 100,000 resources in accessible formats (2006). This is indicative of the quantity of material that could be made available for use by people with vision disabilities, and therefore all others who are for one reason or another not using their eyes as they might to view content.
REVEALWEB's formats are:
Given the size of this collection of well-described, discoverable materials, it is important that any new metadata descriptions are interoperable with this list.
The USA also has a union catalogue maintained by the Library of Congress National Library Service for the Blind and Physically Handicapped (NLS). The Union Catalogue (BPHP) and the file of In-Process Publications (BPHI) can both be searched via the NLS Web site [NLS].
Indicative statistics for the NLS (according to those posted on 2005-01-11) are:
Each year it distributes 23 million books and magazines to a readership of more than 759,000 individuals who cannot read regular print for visual or physical reasons. NLS functions as the largest and frequently only source of recreational and information reading materials and services for a segment of the population who cannot readily use the print materials of public libraries. The NLS International Union Catalog contains 382,000 titles in 22 million copies. (NLS, 2002)
The formats available appear to be press Braille, digital Braille (Web-Braille), audio cassettes, large print text, digital text, maps (tactile), electronic resource, music (Braille), music (large print), and sound recordings (NLS, 2006).
In a fact sheet, NLS explains: "Currently, this service includes the acquisition, production, and distribution of Braille and recorded books and magazines, necessary playback equipment, catalogs and other publications, and publicity and marketing materials" and that, "One of the primary reasons for instituting a national program was to obviate the inevitable difficulty and high cost for individual libraries to acquire books in special formats" (NLS About, 2006). In a sense, this is the same motivation as is being suggested in this thesis for the development of a metadata standard for AccessForAll materials.
NCAM, the National Center for Accessible Media at WGBH in Boston have developed software and techniques for making media of all sorts available to all people. As part of this process, they have developed a clever way of distributing captions and descriptions (known as MOPIX) to theatre and cinema goers. Currently there are more than 300 films available with captions and descriptions (MOPIX, 2006).
etc.....
There are two kinds of content adaptation services: those that adapt the components of a resource to fit a given specification and those that in some way adapt the components, such as converting text into Braille. As well as static, or held content, there are services for creating accessible content - some of which work on-the-fly and others which can be used asynchronously.
The Speech-to-Text Services Network [STSN] has, for some time, been making accessible content alternatives for content that cannot be used by people with hearing disabilities. They describe their three real-time speech-to-text services according to the technology used to process incoming speech:
The STSN has a table that shows differences and similarities among their services. This table also makes clear the sort of services that are valued by people with hearing disabilities. Some of these are relevant in the current context because they represent services that some people will use when they cannot access auditory information.
Steno machine-based Stenography Systems - CART Laptop-based Speed Typing Systems Automatic Speech Recognition Systems (ASR) Verbatim, or near-verbatim translation, i.e., word-for-word Meaning-for-meaning translation, i.e., "all the meaning in fewer words" Communication access usefulness determined by ASR software error rate, reader's error tolerance, skill of speaker, etc. Typist who is trained court reporter Typist who is trained in specific system Trained "Shadow" speaker Info Link CART Info Links TypeWellC-Print Info Links ASR, CaptionMic, iCommunicator, Liberated Learning Initiative (STSN 2006)
As is apparent from the table, human services are provided to render the content accessible to those who are not able to hear it in its original form. Such services exist alongside new ones being developed like those offered and proposed by ubAccess, particularly SWAP that will utilise computers to perform 'intelligent actions' on inaccessible content.
ubAccess has developed a wizard Semantic Web Accessibility Platform SWAP, that can transform a given Web page to have characteristics that will suit users with special needs. As this service depends upon knowing the users' needs, it is appropriate for it to be considered as an example of the type of service that will be enabled by the AccessForAll approach to accessibility.
There are many services that are built into content servers that could be described as adapting content, or components of aggregate content, into suitable composites for users. In general, these are driven by the device and software requirements. The materials delivered to a telephone by a standards compliant browser will at least attempt to adapt the resource for that device. For example, the Opera browser can present the user with a newspaper page in a way that makes sense to someone with a very small screen:
Front page of the Age newspaper on 9/11/2007 in Safari and Opera Mini showing headlines so phone users can easily select what to read or look at.
This is discussed elsewhere?
The early DC accessibility work is relevant because it cleared the way for the AccessForAll approach that has become to main work of that group.
The author's early experience as a professional metadata developer had led to the first DC application profile, and more broadly, the facility that was to be exploited again in the accessibility work. After developing a profile for the Victorian Education Channel, the author developed one for the Victorian Health Channel. This experience was not directly based on metadata about accessibility but the general principles, as espoused by one of the then Directors of the DCMI, Stuart Weibel, were assumed:
New metadata elements are only to be created if they do:
The author therefore suggested working through the existing DC terms to find a combination that would describe the accessibility of resources. Tha aim, at the time, was to be proactive in setting an accessibility agenda for content developers by bringint their attention to the need for accessibility, as much as to provide functional metadata. Some time later, the other Director of DCMI, Eric Millers, strongly defended this position at a DCMI Advisory Committee (as it was) meeting and there was general support for the work.
Over a number of years the following efforts to find a way to define accessibility metadata were promulgated.
The 'rules' for DC metadata have always been that the metadata terms must comply with the Dublin Core model. That the model has not, until late in 2007, been expressed in an unambiguous way has made this process very difficult. Once the accessibility work left the fold of the DC and was led elsewhere based on another type of metadata, the best that coulld be done was to ensure that the new metadata matched as closely as possible the DC model, and that it was at least possible to cross-walk without loss from one system to another.
Given the changing nature of the DC model, there was many iterations of the AfA metadata in an attempt to match the model but they always seemed to fail to do this. Once the model became stable, it was possible to determine the requirements once and for all and the most recent version of the abstract model of the DC AfA metadata appears to do this.
This model and the associated vocabularies have not been formally adopted by DC, which requires the approval of the DC Usage Board, but it has been informally accepted as now matching the rules. Achieving this status required input to the DC process of definition of that abstract model, as well as the development of this one to match it (Pulis & Nevile, 2006).
In 2007, Andy Powell had the following to say in the context of educational metadata:
so what does history teach us? Why are we where we are now? I would argue that the "effort aimed at distilling semantics & simplifying them through delivering sufficient consensus across a significant community of practice" essentially failed. It failed because the approaches reached thru that consensus cost more to implement than the benefits they realise in the context of the original use-case (resource discovery on the Web).
When was the last time you found something because it had been described using DC?
What history tells us is that DC is too complex for the 'simple' resource discovery scenarios envisaged when the initiative started. Those scenarios now tend to be catered for by full-text indexing and social tagging of one form or another. At the same time DC is not complex enough for the scenarios typically found in digital libraries, scholarly communication, elearning, commerce and the like.
Yes, the DCMI Abstract Model tends to move us more towards the latter. Yes, explicitly modelling the entities in the world that we want to describe is more complex than not doing so.
Complex but necessary. All IMHO of course.(Powell, 2007)
In a sense, the metadata being proposed for accessibility is very complex but it is meant to be used differently in different circumstances. The typical use of it is with a single term (Dublin Core or other) where the term is named accessMode and the values identify the perception mode for the content. This information alone will make a huge difference to discoverability for a user. Then, when a resource is made or catalogued by experts and designed to satisfy an accessibility problem, those who have devloped it can use their expertise to give maximum value, and exposure, to the resource.
The final stages of development of the WCAG 2.0 specifications have been under way as the AfA metadata has been finalised as an ISO standard. Convincing the W3C Working Group responsible for WCAG 2.0 to include a requirement for AfA metadata would have made all WCAG 2.0 conformant resources capable for adaptation according to AfA principles. For a number of reasons this was not possible, not the least being that the WCAG authors were not prepared to admit that a resource might be less than conformant and yet 'legitimately' be described by metadata as specified by WCAG. They did consider it important to allow for the use of metadata, however, especially to identify an alternative resource that could be used by a user when that alternative had special feature to make it more useful than a standard, conformant resource, and the original was already WCAG conformant. Given the inclusion of this as a technique, there is, of course, no reason why a developer should not provide full AfA metadata and if there are tools that make this easy, it might happen.