Last updated 17 Nov 2004
Use case 1: Distributed content production, centralised label control
Exemplary Portal Inc. has 40 production centres around the world. Each is responsible for a subdomain of example.org and is largely autonomous. The Exemplary Portal operates two further domains at example.net and example-inc.net for internal functions but these domains are used to supply some content to the public-facing web properties. As well as content produced in-house, Exemplary Portal carries a great deal of third party content.
Although production is spread around the world, corporate liability is concentrated in one department at head office.
Each production centre should arrange for content to carry an identical tag that points to the labelling information. The tag should be regarded as stable over the medium to long term. Labelling information should be under the direct and easy control of the corporate liability department. It is posted online at www.example.org/labels.rdf.
The link tag should therefore be:
This can also be expressed as an HTTP Response Header:
Link: ; /=”/”; rel=”meta” type=”application/rdf+xml”;
This method allows Exemplary Portal to make inclusion of the same link a feature of its standard server configuration.
A client following that link should receive back an RDF graph that is about, or can be interpreted as being about, the resource carrying the link. For example:
ads banners * 1 1 1 1 1 1 1 1 1 0 1 1
[NB. Need to add in domain restriction somehow]
Use case 2: Distributed content production, distributed label control
The Content Management Company is a major portal for a single country, Germany. In accordance with German practice, it categorises its content into age brackets 5: 0-6, 6-12, 12-16, 16-18 and 18+. The Content Management Company defines 5 labels and puts them in an RDF instance.
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
The CMS operated by the Content Management Company uses metadata associated with elements present on each page to assign a label for the generated HTML. This is achieved by inserting one of 5 possible link tags:
Use case 3: Centralised generic labels with local override
This is a combination of use cases 1 and 2. The staff at one of the Exemplary Portal’s 40 production centres feels that a particular page they’ve created should not be associated with the “defaultContentLabel” but should use the “advert” label instead. This is achieved easily by changing the link tag as follows:
The same RDF instance is referenced as for other pages on the example.org site. However, the tag points to a specific fragment of the instance and the filter should respect this rather than processing the application rules.
NB. It is quite possible that a user’s computer will already hold the RDF instance in cache and, in this use case, probably allow the page based on the information it already has. Filters should, however, always follow links to RDF instances to look for data of the type contentLabel. A draft proposal for how a filter should handle multiple labels for a given site is discussed in the label processing document but this is very much open to debate.
Search engine results
A very similar situation obtains in a search engine’s results. The labelling of the results page cannot be pre-determined since, by its nature, it is compiled at runtime based on a user’s input. Similarly, its URL is not a guide to what content is on the page. Use case 3 covers this well since images and other page elements that appear irrespective of the search conducted will be labelled through the centralised system (their URIs are static). The HTML page itself will also be labelled through the centralised system, BUT, if the search engine recognises that the label is inappropriate for the results returned, an extra link tag can be inserted in the HTML that points to a more appropriate label. This will effectively override the ApplicationRules specified in the RDF instance.
| Powered by | ||
 |
 |
|