Revised proposal for RDF-based labels published

Updated 23 February 2005

As part of the Quatro¹ project, and following on from the work of ICRA’s Labelling Working Group², the first technical draft of the proposed new labelling system has been published³.

The new system, based on RDF⁴, offers many advantages over PICS⁵ labels currently used by ICRA and others.

1. Labels can be stored in a separate file with links from content to be labelled, rather than embedding labels within the content itself.
2. As well as the labels themselves, rules for their application can be included in the file. Therefore, if desired, the links from the resources (documents, images, movies etc.) to the label file can be identical, irrespective of which label should be used. The client (filter) has enough information to work out which label to apply to which resource.

   For example, it is possible to write a rule that says “all material on this website should have label A except the following (list) which should have label B. “

3. Whilst it is possible to put rules and labels in a single location and use identical links to point to them, it is also possible to link a resource to a specific label. This offers a PICS-like labelling method and the ability to override rules where present.
4. A single label or set of labels can be applied to any number of websites.
5. Metadata other than content labels can be included in the same system. For example, authorship, publisher and copyright information.
6. The labels and rules in one method offers filters a significant optimization route. Once the labels and rules file (a small text file) is held in the filter’s cache, it can decide whether to retrieve an individual item from the web without first visiting that site or looking it up in a database. All the necessary information is on hand.
7. The system uses the familiar link/rel tags as defined for HTML 4.0. Webmasters are used to these, content management systems support them. This is in contrast to PICS labels that are less familiar and, although fully compliant with all HTML standards, have a syntax that often trips up a CMS or web authoring tool

In order to give an idea of what these advantages mean in practice, some non technical examples are given below.

Non-technical examples

Large scale production content providers

Large scale content providers will typically run many different websites that market their pool of content in different ways. Their servers or content management systems can now be configured to include a standard link to their central label file by default. The same link can be included whatever the domain name, whatever the content type.

Configuring the servers or CMS to include the link is a one time job.

An individual or department can then be given the task – and the tools – to manage the labels. Individual production staff need not be directly concerned with the operation.

ISPs, hosting companies and aggregators

Companies that offer their users web space or an entry in a directory can now easily include labels. The company concerned might define a set of labels, all associated with an age-range or content type. These labels would be held in a single file. When the user signs up for the service, they select which of the labels is appropriate for them and the system then includes a link to the appropriate label in their section of the web. The user doesn’t need to see the ICRA questionnaire or do anything technical at all.

Individual webmasters

It will still be easy to label an individual website in much the same way as is possible now using PICS. A webmaster will visit the ICRA questionnaire, check the boxes that describe his or her content and then add the generated label to their site.

Questions and answers

How much will the first draft change and when?

The system laid out in the draft has been through a good deal of work and review before publication. We are now seeking comment from the wider community with a view to publishing any refinements before the end of January 2005.

ICRA expects to begin working with its members, associate members and others to get the first sites labelled with the new system no later than early February 2005.

How does the Quatro project and trust fit in with this?

The Quatro project¹ is working to create mechanisms through which greater trust can be placed in labels. These are likely to include digital signatures, database look-up, real-time content analysis etc.

The proposal revised in February 2005 includes a mechanism for specifying who created the labels. This is designed to be more than just informational. It will provide a location at which information can be made available about how the trust a user places in the label can be enhanced.

When will ICRA fully move across to the new system?

ICRA will continue to recognise, and encourage others to recognise, existing PICS labels for the foreseeable future. Work is now starting on creating the tools to generate, read and act on the new labels to work alongside the existing PICS systems. A new reference module for filters is now being commissioned that will be integrated with ICRAplus and will also be available for use in other systems.

Once the tools have been built, ICRA will switch over to generating new system labels only.

Does this tie in with the new ICRA vocabulary?

Yes. All labels generated using the new technical platform will use the revised vocabulary⁶. The English language version is already available as an RDF schema⁷.

Does the new system only work with ICRA labels?

No, absolutely not. The new system will support any labelling scheme. ICRA is simply an example of how it can be used.

What about labelling digital content other than web pages?

The proposed system allows a content provider to make statements like “label A applies throughout but there are occasional scenes described by label B and a single scene described by label C.” This is exemplified in a use case added in the February revision.

Links

Powered by