This paper presents the outcomes to date of the annotation interoperability component
of the Open Annotation Collaboration (OAC) Project.
Annotating is both a core and pervasive practice for
humanities scholarship. It is used to organize, create and share knowledge.
Individual scholars use it when reading, as an aid to memory, to add commentary,
and to classify documents. It can facilitate shared editing, scholarly
collaboration, and pedagogy. Although there exists a plethora of annotation
clients for humanities scholars to use (Hunter
2009) - many of these tools are designed for specific collection
types, user requirements, disciplinary application or individual, desktop use.
Scholars are also confronted with having to learn different annotation clients
for different content repositories, have no easy way to integrate annotations
made on different systems or created by colleagues using other tools, and are
often limited to simplistic and constrained models of annotations. For example,
many existing tools only support the simplistic model in which the annotation
content comprises a brief unformatted piece of text. Many tools conflate the
storage of the annotations and the target being annotated.
Frameworks for annotation reference are inconsistent, not coordinated, and
frequently idiosyncratic, and the constituent elements of annotations are not
exposed to the Web as discrete addressable resources, making annotations
difficult to discover and re-use. The lack of robust interoperable tools for
annotating across heterogeneous repositories of digital content and difficulties
sharing or migrating annotation records between users and clients – are
hindering the exploitation of digital resources by humanities scholars. Hence
the goals of the Open Annotations Collaboration (OAC) are:
In the remainder of this paper we describe related efforts that have informed the
development of our Annotation Data Model. We then describe the data model itself
that lays a foundation for follow-on work involving demonstrations and reference
implementations that exploit real-world repositories such as JSTOR, Flickr Commons, and MONK and leverage
existing scholarly annotation applications such Zotero, Pliny and
Co-Annotea.
Despite the vast body of work regarding annotation practice, annotation models,
and annotation systems, little attention has been paid to interoperable
annotation environments. The few efforts in this realm to date comprise:
An analysis of these existing models reveals that on the whole, they have not
been designed as Web-centric and resource-centric, or that they have modeling
shortcomings that prevent any existing resource from being the content or target
of an annotation and from giving an annotation independent status as a resource
itself. Further requirements that we have identified that these approaches fail
to fully support include:
By exploiting the Web- and Resource-centric approach to modelling annotations, we leverage existing standards and facilitate the interoperability of annotation applications. In the OAC model, an Annotation is an Event initiated at a date/time by an author (human or software agent). Other entities involved in the event are the Content of the Annotation (aka Source) and the Target of the Annotation. The model assumes that the core entities (Annotation, Content and Target) are independent Web resources that are URI-addressable. This approach simplifies and decouples implementation from the repository. An essential aspect of an annotation is the (implicit or explicit) expression of “annotates” relationship between the Content and the Target. The model allows for Content and Target of any media type and the Annotation, Content, and Target can all have different authorship. In situations where the annotation Content or Target is a segment or fragment of a resource (e.g., region of an image), we will draw on the work of the W3C Media Fragments Working Group to specify the fragment address. Figure 1 illustrates the alpha version of the OAC data model.
In order to evaluate and demonstrate the feasibility of the OAC Data Model, an
initial set of use cases has been developed that are representative of a range
of common scholarly practices involving annotation. This preliminary set is
available from the OAC Wiki as OAC User Narratives/Use Cases
For example, Figure 2 illustrates a scholarly annotation example involving multiple targets, in which a scholar is making a comment on the differences between segments in scholarly editions of the poem “The Creek of the Four Graves” by Charles Harpur.
Figure 3 below illustrates the corresponding OAC model for the use case in Figure 2 in which a single annotation Content applies to two Target resources.
The proposed OAC Data Model will enable the sharing and discovery of annotations beyond the boundaries of individual solutions or content collections, and hence will allow for the emergence of value-added cross-environment annotation services. It will also facilitate the implementation of advanced end-user annotation services targeted at humanities scholars that are capable of operating across a broad range of both scholarly and general collections. Furthermore, it will enable customization of annotation services for specific scholarly communities, without reducing interoperability. The proposed work will also enable more robust machine-to-machine interactions and automated analysis, aggregation and reasoning over distributed annotations and annotated resources. By grounding our work in a thorough understanding of Web-centric interoperability and embedded models implemented by existing digital annotation tools and services, we create an interoperable annotation environment that will allow scholars and tool-builders to leverage prior tool development work and traditional models of scholarly annotation, while simultaneously enabling the evolution of these models and tools to make the most of the potential offered by the Web environment.
Acknowledgments
The Open Annotations Collaboration (OAC)
is funded by the Andrew W. Mellon Foundation. The authors would also like to
acknowledge the valuable contributions to this work made by: Neil Fraistat, Doug
Reside, Daniel Cohen, John Burns, Tom Habing, Clare Llewellyn, Carole Palmer,
Allen Renear, Bernhard Haslhofer, Ray Larsen, Cliff Lynch and Michael Nelson.
Figure 2 is courtesy of Anna Gerber, Senior Software Engineer on the Aus-e-Lit
project.