Detecting CC Attribution License Violations with Flickr Images on the Web

Oshani Seneviratne

Creative Commons Tech Summit
12 December 2008

Overview

• Motivation
• Creative Commons Licenses in Flickr
• System Design
• Future Directions

Overview

• Motivation
• Creative Commons Licenses in Flickr
• System Design
• Future Directions

Why do we need to detect CC license violations?

• To verify your own work
• Notify when somebody violates your CC licenses

Related Work

• Commercial Image Trackers
  • PicScout
  • Attributor
• License Detection Tools
  • Flickr Search Interface
  • MozCC
  • Operator
• License Embedding Tools
  • ThinkFree (web based commercial office suite)
  • License Tagger
  • Flickr Image Reuse for OpenOffice

Overview

• Motivation
• Creative Commons Licenses in Flickr
• System Design
• Future Directions

Creative Commons Attribution

• Human Readable Form:
  

  "You let people copy, distribute, display, perform, and remix your
  copyrighted work, as long as they give you credit the way you request"

• Machine Readable Form:

  @prefix xhtml: <http://www.w3.org/1999/xhtml/vocab#>
  @prefix cc: <http://creativecommons.org/ns#>
  
  <http://flickr.com/photos/oshanis/random_number>
     xhtml:license
     <http://creativecommons.org/licences/by/2.0> ;
     cc:attributionURL
     <http://flickr.com/photos/oshanis> ;
     cc:attributionName
     "Oshani Seneviratne" .
  

You are expected to do something like...

Photo by Oshani Seneviratne, BY-CC-2.0.

Flickr

• Popular photo sharing site
• Has a nice API
• Users can express their rights as follows:

Creative Commons Licenses in Flickr

• Still on Creative Commons 2.0!
• And only the following set of licenses are supported:
  
  
  
  

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

Overview

• Use Cases
• Creative Commons Licenses in Flickr
• System Design
• Future Directions

System Design

Output

Current Issues

• Does not work if the images are downloaded from Flickr and used. Because the validator only works with images with the following URI format:
  

  http://farm{farm-id}.static.flickr.com/{server-id}/{id}_{secret}.(jpg|gif|png)
  OR
  http://farm{farm-id}.static.flickr.com/{server-id}/{id}_{secret}_[mstb].(jpg|gif|png)
  OR
  http://farm{farm-id}.static.flickr.com/{server-id}/{id}_{o-secret}_o.(jpg|gif|png)
  

• Although Flickr provides support for EXIF, and we can write license information as EXIF data, it could be easily overwritten.

Overview

• Use Cases
• Creative Commons Licenses in Flickr
• System Design - With Semantics
• Future Directions

Scenario

Outcome with Explanations

Model in AIR Rule Language

@prefix cc: <http://creativecommons.org/licenses/by/3.0>.
@prefix xhtml: <http://www.w3.org/1999/xhtml/vocab#>.
@prefix dc: <http://purl.org/dc/elements/1.1>.
@prefix air: <http://dig.csail.mit.edu/TAMI/2007/amord/air#>.

@forAll :EVENT, :P1, :P2, :WORK, :LICENSE, :DERIVATIVE. 
 
:CC_BY_Policy a air:Policy;
   air:rule [
      air:pattern {
         :EVENT a air:UseEvent;
             cc:work :WORK.
         :P1 a foaf:Person.
         :WORK dc:creator :P1;
             xhtml:license :LICENSE.
         :LICENSE cc:requires cc:AttributionName. };
    air:rule [
      air:pattern { 
	        :DERIVATIVE cc:derivativework
	        :WORK;
	        dc:creator :P2 . };
          air:rule [
              air:pattern { 
                :DERIVATIVE cc:attributionName :P1 . };
             air:assert { 
                :EVENT air:compliant-with 
                :CC_BY_Policy. };
             air:alt [
                air:assert { 
                :EVENT air:non-compliant-with 
                :CC_BY_Policy. };
            ];
         ];
      ];
  ].

Overview

• Use Cases
• Creative Commons Licenses in Flickr
• System Design
• Future Directions

Extend to Other Licenses

• Non-commercial Use
• What is the definition of a commercial use?
• Can an automated tool ever figure out a "Fair Use"?
• Share Alike
• Check the RDFa of the embedded image with the original.

Extend to Other Media

• Video - For e.g: YouTube
• No CC field unlike in Flickr
• Therefore no standard method to query license data, other than by scraping the video description or the comments
• Documents
• Need to crawl the entire WWW to obtain CC license data!

Questions?

oshani@csail.mit.edu