Web Posted on: August 24, 1998

Interactive Development of WWW Sites Based on Meta-data Collection

Filip Evenepoel
Katholieke Universiteit Leuven
Res.Group TEO –
Document Architectures
Kardinaal Mercierlaan 94, B-3001 Leuven – Heverlee, Belgium
tel: + 32 16 32 18 66, fax: + 32 16 32 19 86,
email: Filip.Evenepoel@esat.kuleuven.ac.be

Keith Gladstone
Royal National Institute for the Blind
IT Systems and Services Development
Bakewell Road, Orton Southgate, Peterborough, PE2 6XU, UK
tel: +44 1733 375 169, fax: +44 1733 375 255,
email: KGladstone@rnib.org.uk

Jan Engelen
Katholieke Universiteit Leuven, Res.Group TEO –
Document Architectures
Kardinaal Mercierlaan 94, B-3001 Leuven – Heverlee, Belgium
tel: + 32 16 32 18 66, fax: + 32 16 32 19 86,
Filip.Evenepoel@esat.kuleuven.ac.be

1. Summary

In this paper, an outline of the on-line dissemination activities of the ARTNet Support Action (Telematics Application Programme, project SU 9111) will be given. We will indicate how information on the actual data can be used to enhance the information dissemination process.

| Top |

2. Brief Introduction to the ARTNet Support Action

In the field of Assistive and Rehabilitation Technology, dissemination of information is an important activity. This particular domain is relatively small compared to other main stream activities, resulting in disperse, often local, activities known to few people. Through a co-ordinated effort in the organisation of information dissemination, the ARTNet Consortium hopes to contribute to the wider spread of information and eventually to an improved use of the sparse resources available in this sector. The medium we have chosen for our dissemination purposes is the World Wide Web.

Before the start of this project, one could find very little European information on this subject on the WWW, compared with the large number of ongoing activities. However, we can expect that the organisations active in this domain must have a lot of public information they might wish to share with others. Compared with the situation in North America, where co-ordination efforts have been successful, Europe was certainly lagging behind. Our intention in the first instance was to at least provide a dissemination platform on the Web for Europe through a single point of access (a place to start from).

As will be outlined in the remainder of this paper, we believe that more can be accomplished through the full exploitation of all the possibilities offered on the Web (and other Internet applications). In order to improve the quality of the provided information to the users of the ARTNet dissemination platform, we have and are continuously updating the methods for storing, cataloguing and presenting the information. Besides the actual collection of information, we are also preparing for technology transfer related to the use and set-up of tools for WWW applications.

| Top |

3. Outline of the Methodology

Providing large quantities of information to users, without a information management system increases the risk of information overload and will not provide the desired effect. The core theme in any dissemination effort (and as such in the accompanying application) should be the information management system. In the framework of the ARTNet Support Action, we are developing a mechanism which will allow us to provide the desired information to a user of the system in an as generic way as possible, independently of that user’s capabilities and limitations. To this end, we are interested in information on the data which needs to be disseminated, also called meta-data (information about the information).

3.1 Meta-data Specification

It was recognised in the early stages of the project that classification of information would be necessary. This classification of the ARTNet documents, which will eventually be provided through the World Wide Web, is a necessary task to enable logical structuring of these documents. The usability and accessibility of the documents will be highly dependent on the structuring which can be provided through these systems.

The collection of the different characteristics on which the logical structure will be developed, form a multidimensional space which contains the following collections of characteristics: the CORE study results [Reference 1]; the ISO 9999 classification of technical aids; a topic list; language and geographical relevance; type of disability relevant for the document and its intended audience.

The development of this multidimensional space is the result of the co-operative work of the different partners in ARTNet, both through their insights in this subject area and through the evaluation and study of different existing sites on the Web. It is important to point out that the collection of characteristics need to be dynamic, in order to allow for growth and evolution of the system. An in-depth review of the meta-data specification as it is used for ARTNet can be found in [Reference 2].

3.2 Fuzzy Nature of Meta-data

All the characterisation schemes proposed in ARTNet, except the language and geographic relevance, can be thought of as being fuzzy. To explain the fuzziness of meta-data, we will use the following mathematical terms: dimension and set. Every characterisation scheme (CORE, ISO, topics, …) which can be applied to a document forms a dimension, while every dimension is composed of sets, formed by the items (characteristics) in each characterisation scheme (i.e. each characteristic is a set).

For example, the dimension "CORE subjects" contains eight sets of documents (which are the elements in the sets): research, development, production, trade, procurement/financing, service delivery, usage and legislation.

Every document can be placed in every dimension were it can be an element of one or more sets. This means that for every document, we will be able to assign at least one characteristic item to it in every scheme.

Let us assume for a moment that we are working with ordinary sets (not fuzzy). A document can then be either an element in one set, or it can be an element in more sets, which will then make it an element in the intersection of these particular sets. Based on this assumption, it is possible for a document to be relevant in different domains in a scheme (for example relevant for end-users and carers in the intended audience scheme), but it is impossible to assign a weight to any particular characteristic.

The possibility to assign a weight to a characteristic (a set) in a characterisation scheme (a dimension) is important to increase the usability of the collection of documents. The need to maximise the usability of the ARTNet Web sites drives us towards using the fuzzy nature of the sets.

For the purpose of the ARTNet dissemination platform, we will use a 101 point scale (percentages) to indicate the relevance of a document for a particular set. We will capitalise on the fuzzy nature to increase the search capabilities of a user on the ARTNet sites.

| Top |

4. The Document Life Cycle as Exploited in ARTNet

Following the document life cycle, we will discuss how the meta-data on the documents is first collected and normalised (part of the authoring process), then stored and finally used for the automatic generation of topic based lists (publishing) and simple document navigation (end-user browsing).

4.1 Document Creation

The most important tool for an author will be the document editor used to create the WWW pages in HTML. The interface between the author and the ARTNet dissemination platform is organised through the WWW itself. Hence, the author will need to have access to a WWW browser which is capable of handling forms (most current packages). Supporting tools for document authoring, and information registering will be organised through (protected) WWW pages on a central server.

As meta-data is used in the different steps of our document publishing system, we have invested a lot of effort in the development of tools to automatically generate the meta-data. Based on the properties of the HyperText Markup Language (HTML, the lingua franca of the WWW), we create a document template in which the required information is automatically stored based on a questionnaire completed by the author.

The major part of the collected meta-data are a set of topic characteristics (which form a multidimensional fuzzy space), which characterise the document based on pre-defined keywords. These document characteristics, along with other data such as the author, date of creation, revision number and so on, are stored in a database from which we will be able on one hand to create topic based document lists, and on the other hand to allow topic based searching by end-users.

4.2 Document Verification

A particular aspect of the ARTNet solution lies in the special requirements to enforce the accessibility of the documents for users with a disability. As an additional aid in the authoring process, a publishable document is checked on the correctness towards the allowed document structures (more strict than regular HTML), on the specific features for accessibility (in particular, but not exclusively for visually impaired people) and on the availability of the earlier created meta-data. Once all these steps have been passed, a document is ready for publication.

Although verification will mostly be an automated process, it has a serious impact on the document design and creation. In order to support the author, a set of guidelines [Reference 3] have been collected and synthesised for the ARTNet platform. Starting from the guidelines, an author can create documents so that uniformity, accessibility and usability are safeguarded. The next step, to be undertaken by the author, will be the registration of the document in the ARTNet repository. Again, the interface to the author will be an HTML form.

Since we want to limit the authors in their possibilities for creating documents, a minimal set of validation steps are required. A first validation, mainly to check uniformity, will check the use of the HTML language, and the limitations imposed by ARTNet. As the meta-data generation in the document creation cycle is optional and the information can still be changed by the author, it will be necessary to validate the correctness of the supplied meta-data. The submission time will also be used to validate the links which can be included in the documents. These last two aspects are related to the usability of the information.

A final step in the validation exercise is the accessibility of the information. Currently, it is unclear if such a test needs to be developed specifically for the ARTNet repository, or if an existing tool (such as for example Bobby [Reference 4]) can be used.

After the validation exercise, the document will be included in a database with all other ARTNet documents. As of that time, the document will be included in the topic based lists, the search pool, and all other ARTNet specific applications.

4.3 Document Distribution

The major advantage of using the World Wide Web is the possibility to create hypertext, a collection of linked information chunks. Since the World Wide Web is a non-controlled system, the burden of checking whether the address of a link-end, used in a document, is a task of the owner of the page. Checking the integrity of the links is certainly part of the document verification process, but is remains a potential problem through the distribution stage as well. The only solution is to run at regular times a link verification process (also called a robot).

The way in which we store the information on the documents (a database) allows us to automatically generate document lists based on the topics as provided by the authors. Based on the same principle, we will be able to provide better answers to user queries, based on knowledge on the user (type of user, type of disability, specific preferences) if this information is available. Eventually, this will result in better user satisfaction over currently available search engines on the WWW.

| Top |

5. Conclusion

During the entire specification and design cycle of the ARTNet dissemination platform, we have and are continuously trying to marry the best of both worlds. One the one hand, we wish to use as much as possible off-the-shelf solutions, used in main stream applications, to deliver information, and on the other hand, we need to provide accessible and usable information to our audience. Through combining these efforts, we expect to obtain a platform where people get the information they desire in an efficient manner.

| Top |

References

1. Consensus creation and awareness for R&D activities in technology for disabled and elderly people (CORE), Deliverable 2, TIDE projects 126/213.
2. The ARTNet WWW Server Set-up, Filip Evenepoel et al., Deliverable 3.1 of the ARTNet Support Action (can be obtained from the authors).
3. ARTNet WWW Guidelines, Keith Gladstone et al., Deliverable 4.2 of the ARTNet Support Action (can be obtained from the authors).
4. Bobby Web Accessibility Tool, Center for Applied Special Technology, URL: http://www.cast.org/bobby/

| Top | | TIDE 98 Papers |