WSRI Exchange : Detecting CC License Violations with Flickr Images on the Web

Oshani Seneviratne

29 August 2008

Overview

• Creative Commons Licenses
• System Design
• Algorithm
• Current Issues
• Future Directions

Creative Commons Licenses

Attribution
Attribution-NoDerivs
Attribution-NonCommercial-NoDerivs
Attribution-NonCommercial
Attribution-NonCommercial-ShareAlike
Attribution-ShareAlike

Attribution

Creative Commons Legal Code defines "Attribution" to be something like:

Photo by Oshani Seneviratne, BY-CC-2.0.

System Design

Algorithm

• Needs a seed site/blog to crawl
• Does BFS on the seed site
• Find out embedded flickr photos there
• Use the Flickr API to get all the information related to the photo using the photo id
• Use the QDOS SPARQL endpoint to obtain the photo owner's data (FOAF URI, etc)
• Check for license attribution on the site with all of the owner's data

Current Issues

• Does not work if the images are downloaded from Flickr and embedded in the site
• Can write license information as EXIF data, but it could be overwritten
• BFS is only limited to all the links within the seed site
• Locality of the search for creator within the web page should be improved
• Notification to user for any license terms violation is not implemented. Needs user consent?

Future Plans

• Extend to other CC licenses
• Extend to other data usage scenarios (for e.g. YouTube)
• Track provenance of images using metadata (instead of relying on the URIs)
• Social Verification: i.e. use the FOAF graph to control access for viewing, tagging and commenting on photo sharing sites
• Automatically inject the attribution details whenever a photo is linked

Demo

Questions?

oshani@csail.mit.edu

Appendix

Creative Commons Schema

From the RDFa schema defined at http://creativecommons.org/schema.rdf

Extended Creative Commons Schema

Schema in N3

Scenario in AIR (compliant)

Scenario in AIR (non-compliant)