We thank the reviewers for their valuable feedback, and hope the following responses will answer their questions.

Review 1
========

Q1.1:
"This paper has demonstrated several tools that lay the foundation for such policy aware systems" is a bit exaggerated.

A1.1:
We agree that the use of the word "foundation" is too strong. Even though there are services such as 'Attributor' (cited as [1]), we haven't come across any open framework that lets users to validate their works against license violations. The concept of the Semantic Clipboard has been tried before (cited as [16]), but our approach is different and can be used for different kinds of media. 
Perhaps the statement "This paper has demonstrated several tools that enable the development of such policy aware systems" would have been more appropriate.

Q1.2:
What happens when users are not cooperative?

A1.2:
We believe this is another research question by itself. The reason for exclusion of this from the discussion is because we are only dealing with making users policy aware (hence the title "Policy Aware Content Reuse"), and NOT policy enforcement. But we believe that a brief mention of this would have been appropriate. 

Q1.3:
What part of this is actually "Semantic" Web oriented?

A1.3:
The Semantic Clipboard looks for and extracts license metadata expressed in RDFa and stores them in an RDF License store. This process is explained in detail in Section 4.2  We understand that RDF is the foundation of the Semantic Web and thus claim that tool is Semantic Web oriented.

Q1.4:
Do we need more intelligent tools with reasoning capabilities to detect complicated license infringements, etc?

A1.4:
This is a very good point. We have mentioned the requirement of detecting complicated license infringements in Section 6's para 1 and 2 where the need to detect license violations other than attribution license violations (i.e. commercial use, derivative use, etc) are highlighted.

Review 2
========

Q2.1: 
There is a conceptual jump between 2.2 and 3.1.

A2.1:
Section 3.1 focusses on the experiment which checks for CC licenses (discussed in 2.1) which are expressed inline with content (discussed in 2.2). At the time of writing the paper we did not notice a big jump between the two sections, but agree that we could've provided a much smoother transition.

Q2.2:
Section 3.1 needs more detail on how the analysis was done.

A2.2:
Space limit prevented us from elaborating. But we will try to add more details about the analysis to the paper.
Here are the answers to the reviewer's specific questions:
* The sampling method was 'simple random sampling' on clusters of sites gathered during a particular time frame.
* The number of sites are small because we did a manual inspection to see if there were any errors (false positives and false negatives.)
* Regarding the comment - "I would presume the authors checked for correlations within sites': We manually inspected if the different samples contain the same sites. Our inspection proved that the correlation was very minimum.

Q2.3:
The section in 4.2 entitled Data Purpose Algebra Analogy doesn't work for me ... Why is this section here? ... the authors should explain what this particular algebra will solve over other methods.

A2.3:
We tried to explain the inner-workings of the Semantic Clipboard in terms of a formal algebraic model. This model forms a basis for encoding attribution XHTML using the license metadata expressed in RDF. However, we could drop this section and use that space to provide details of the experimental setup and analysis.

Q2.4:
Does Qd(i) = Ad(i)?

A2.4:
No. Ad(i) is the user agent that acts on the data item i.

Review 3
========

Q3.1:
On the negative side, the experimental setting is not described very clearly ... the section on refining the experiment after noticing false positive is hardly comprehensible

A3.1:
Apologies for making the description related to issues and refinement concise due to space limitations. However, we've outlined several issues very clearly: 1. Not being license compliant when it's your own photo 2. Low adoption of ccREL on the Web and 3. Services such as Tumble-logs cutting down text that otherwise would indicate that the user was actually being license compliant. 
See also A2.2.

Q3.2:
My suggestion is here to clarify how the algebra is used at runtime by the Semantic Clipboard, and why it cannot be used to detect conflicts by adding one parameter. Maybe is it better to use an OWL ontology for that purpose?

A3.2:
Yes we could. Complex reasoning in the Semantic Clipboard is a feasible future direction of this project. Thanks for the suggestion.
See also A1.4.

Q3.3:
Regarding references to: 'German SmartWeb project' and 'EU Metokis project'

A3.3:
Indeed worth referring to these projects. Our work differs from these projects because we do not require OWL representations of content, but rather the license descriptions in RDFa that can be easily extracted from our tools.