3 Notation

Examples throughout the document will be conveyed as both a diagram and in the Turtle RDF format [TURTLE]. The Turtle examples do not provide namespace declarations, and should be considered following these namespaces:

The diagrams of the examples use the following style:

• Instances are depicted as colored ellipses
  • Instances with a resolvable URI have a single line border
  • Instances with a non-resolvable URN have a double line border
• Classes are depicted as white rectanges
• Literals are depicted as white lozenges
• Relationships are depicted as straight, black lines (predicates where the range is a Resource)
• Properties are depicted as curved, black lines (predicates where the range is a Literal)
• Class instantiation (rdf:type) is depicted as a straight black line with white arrow head
• Example instance identifiers or example literal values always end in a number (eg 'anno1' is a specific instance of an Annotation, whereas 'oa:Annotation' is a real class)
• Relationships not explicit in the model, but important for understanding, are depicted as curved, dashed, colored lines
• Resource boundaries not explicit in the model, but important for understanding, are depicted as grey dashed boxes around the components.

4 First example: Bookmarking a Webpage

In order to rapresent a Bookmark of the W3C Open Annotation Community Group homepage, we start by creating an instance of the oa:Annotation class; then we link such instance to the webpage (Target) identified by the homepage URI http://www.w3.org/community/openannotation/. Adding the Target content type and the dc:format (dctypes:Text and "text/html"), can be useful for applications consuming the annotation.

Figure 6. Bookmark of a webpage, step 1, defining Annotation and Target

We can then provide the oa:Motivation that encodes the reason(s) why the Annotation was created. In this specific case, the Motivation is oa:bookmarking:

Figure 7. Bookmark of a webpage, step 2, adding the Motivation

Next, we are going to provide the Provenance information. The Open Annotation model allows to encode both the agent that created the annotation and the agent that took care of the serialization of the annotation.

Figure 8. Bookmark of a webpage, step 3, adding the Provenance information

In order to improve human readability, we can add a textual description as a Body.

5 Second example: Tagging an Image

We are now going to create multiple tags for the Wikipedia image of the Eiffel Tower here identified by the URI http://alturl.com/wxidq.we start by creating an instance of the oa:Annotation class; then we link such instance to the image (Target) identified by the URI http://alturl.com/wxidq. Adding the Target content type and the dc:format (dctypes:Image and "image/jpeg"), can be useful for applications consuming the annotation.

Figure 10. Tagging an image, step 1, defining Annotation and Target

We can then provide the oa:Motivation that encodes the reason(s) why the Annotation was created. In this specific case, the Motivation is oa:tagging:

Figure 11. Tagging an image, step 2, adding a Motivation

In order to illustrate the different tagging mechanisms supported by the Open Annotation data model we are going to create three different tags:

• A textual tag: "Eiffel Tower". This is the embodiment of the 'classic' folksonomic approach. In this example we are using an embedded textual body (See specs section 2.1.2 Embedded Textual Bodies). The textual body is here identified by a UUID URI.

Figure 12. Tagging an image, step 3, adding a free text tag

• A semantic tag through a non-informational resource: DBpedia entry for the Eiffel Tower. This is the pure semantic tag construct performed by linking to terms coming from ontologies and vocabularies and by typing such terms as oa:SemanticTag.

Figure 13. Tagging an image, step 4, adding a semantic tag with a non-informative resource

• A semantic tag through a webpage: Wikipedia page about the Eiffel Tower. This is a non ideal case that however is common scenario. Many applications use the URis of webpages to identify concepts. For instance I could have a webpage about a particular entity in biomedicine and, without having an ontology or terminology for that, I use that URI also to indicate the concept. As the same URI is obviously identifying the webpage, we can treat it as a pure non-informational resource. We basically define a new URI that is 'a oa:SemanticTag that corresponds to a foaf:page".

Figure 14. Tagging an image, step 5, adding a webpage as a semantic tag

As already mentioned in section 2, the Open Annotation model allows for multiple bodies. Therefore we can now pull together the three different ways of encoding tags in one single annotation. This is just an example, normally you would probably not create three different tags tags for the same entity, unless the goal is to infer mappings.

Figure 15. Tagging an image, step 6, pulling multiple tags together.

Finally, as in the previous example, we can add provenance

Figure 16. Tagging an image, step 7, adding provenance.

The above example in Turtle RDF format:

 ex:anno a oa:Annotation ;
   oa:hasTarget <http://alturl.com/xxbxn> ;
   oa:hasBody ex:uuid ;
   oa:hasBody ex:semtag1 ;
   oa:hasBody <http://dbpedia.org/resource/Eiffel_Tower>;
   oa:motivatedBy oa:tagging ;
   oa:annotatedBy ex:person1 ;
   oa:annotatedAt "2012-02-12T15:02:14Z" ;
   oa:serializedBy ex:software1 ;
   oa:serializedAt "2012-02-12T15:02:14Z" .
 
 <http://alturl.com/xxbxn> a dctypes:Image
   dc:format "image/jpeg" .
 
 ex:uuid a cnt:ContentAsText ;
   cnt:chars "Eiffel Tower" ;
   dc:format "text/plain" ;
   cnt:characterEncoding "utf-8" .

 ex:semtag1 a oa:SemanticTag ;
   foaf:page <http://en.wikipedia.org/wiki/Eiffel_Tower> .

 <http://dbpedia.org/resource/Eiffel_Tower> a oa:SemanticTag.

 ex:person1 a foaf:Person ;
   foaf:mbox <mailto:john.doe@example.org> ;
   foaf:name "John Doe" .
 
 ex:software1 a foaf:Agent, prov:SoftwareAgent ;
   foaf:name "ExAnnotator" .

5 Third example: Commenting a text fragment within a Webpage

In this example we will select a text fragment in the description displayed in the homepage of the Open Annotation Community Grooup and we will comment on it. (see Figure 17)

Annotating fragments of resources is a very common use case for the Open Annotation model. In this example we will create a multiple-bodies annotation that targets a specific text fragment in a Webpage. The Open Annotation model allows identifying resources fragments with both Fragment URis (see 2.1.4 Fragment URIs Identifying Body or Target) or with the SpecificResource (see 3.1 Specifiers and Specific Resources) and Selector (see 3.2 Selectors) mechanism.

Figure 17. Commenting on a text fragment, step 1, identifying the fragment within the Open Annotation Community Group homepage.

We can start by creating the instance of the oa:SpecifResource that is going to link to the original entire resource (Open Annotation Community Group homepage) through the relationship oa:hasSource.

Figure 18. Commenting on a text fragment, step 2, identifying the Specific Resoruce source.

As I am interested in creating an annotation that could potentially display on both the HTML and the PDF version of the same document, I will opt for a Text Quote Selector (see 3.2.2.2 Text Quote Selector). The idea of such selector is simple: we identify the match by defining the text preceeding the match (oa:prefix), the match itself (oa:exact) and the text following the match (oa:suffix).

Figure 19. Commenting on a text fragment, step 3, defining the selector (the exact match has been shortened).

Figure 20. Commenting on a text fragment, step 4, adding body and provenance (the exact match part of the selector has been shortened).

Fourth Example: Annotating an image webpage with a video

The Wikipedia page about the 'Hubble Deep Field' contains an image (see figure 21).

Figure 21. Wikipedia page about 'Hubble deep field' that contains an image identified by the red arrow.

A user creates a video on YouTube that discusses the Hubble Deep Field Image. The video is therefore the Body of the Annotation, and the image is the Target, as the video is about the image. We start by defining the target of our annotation:

Figure 22. Defining the image as the annotation target.

The Open Annotation model gives the option of recording the webpage that the user was looking at while performing the annotation through the use of a SpecificTarget and the relationship hasScope.

Figure 23. Defining the image as the annotation target within the Wikipedia page (scope).

We will now define the body of the annotation as the YouTube video.

Figure 24. Defining the video as the annotation body.

Figure 25. Adding provenance.

Figure 26. Identifying a Body fragment throuh a Media Fragment URI [Media-Frag].