Metadata You Can trust?

The Web Content Label Incubator Activity (WCL-XG)

Phil Archer
ICRA

Use case 1

The Exemplary Multimedia Company offers a range of ring tones, video clips, full TV programmes, images and text. In order to maximize their assets they make metadata available that describes each resource in terms of:

content type, subject matter, authorship, genre etc.
compliance with Mobile Web Initiative Best Practice (mobileOK)
compliance with WAI guidelines
presence or absence of nudity, sexual content, violence etc.

Use case 1

A lot of the metadata for these resources will be the same

All content meets WAI AA and mobileOK standards, and is copyright Exemplary Multimedia Company.
There is no sex or violence in any content but resources whose URLs contain the word "-pg" may portray bare breasts, bare buttocks, alcohol or gambling.
Some metadata is unique to a given resource, such as title and author. This can be accessed using a URI associated with the resource. This might be a URL, an internal ID number or the resource's ISAN number.

The point of use case 1 being...

The Exemplary Content Aggregator receives the metadata as a single resource
The commercial relationship between the content provider and content aggregator means the data can be taken at face value.
For security, however, the data is digitally signed by the Exemplary Multimedia Company
Customized content can be presented to the end user: mobile, desktop, IPTV channel

Use case 2: Trustmarks

(Think TRUSTe, BBBonline, Medical & scientific website labelling)

Reviewed site includes a link/rel tag pointing to some RDF data
Agents can follow the link and get the data relevant to the labelled site
Database can be exported wholesale to people like search engines and aggrgators
Data comes from a third party and so is as trustworthy as that thrid party

Use case 3: Multiple labels

A website may carry several labels
- Some digitally signed
- Other metadata
Adds descriptions using widely recognised vocabulary
- Well defined vocabulary allows assessment by machine learning

The point being...

If you trust the digitally signed data and you can analyse the content in terms of the description, you can probably make use of the unsigned, un-analysed, self-declared metadata too.

Use case 4: RSS, Atom etc

Film review site offers reviews of family films, summarised in both RSS and ATOM feeds
Most reviewed films have an MPAA rating of G and/or BBFC rating of U, declared at feed level
However, some films are rated PG-13 or 12 which is declared at the item level and overrides the default metadata

Key elements

Re-use of a small amount of metadata to do a lot
Default and override
Trust as part of the model from day 1
Designed to make the addition of metadata easy and commercially desirable
Applicable to anything on the internet, including IPTV, movies, games etc.

How do we do that?

That's what the XG is for...

... but we think we have a pretty good idea already.

Basic idea

Link/Rel=meta from resource to an RDF Instance - can be in XHTML or HTTP. Others?
XML Module?</>
RDF Instance includes the actual label (an RDF class with a bunch of properties)
AND simple patterns against which you apply the resource's URI to identify the correct label
Mean URI, not URL - think ISBN, ISAN etc