Here are the ways I can think of that an automated system could know that a web page is licensed:

1. has a link to a known licence URL
2. has a link that has rel="license" attribute in the tag, and a legal expert confirms that the link target is a licence URL
   
3. has a meta tag with name="dc:rights" content="URL", and an expert confirms that the URL is a licence
   
4. has embedded or external RDF+XML with license rdf:resource="URL"
5. natural language, such as "This web page is licensed with a Creative Commons Attribution 1.0 Australia License"
6. system is told by someone it trusts
   

Here are the ways I can think of that an automated system could find new, previously undiscovered, types of licences (or at least URLs thereof):

1. URL is in rel="license" link tag, expert confirms
   
2. URL is in meta name="dc:rights" tag, expert confirms
   
3. URL is in RDF license tag
4. page contains an exact copy of a known licence
   
5. system is told by someone it trusts

If you can think of any other items for either of these lists, please let me know.

Labels: ben, quantification