JISC

Overview

Tagging is an increasingly popular mechanism used on the web to add personal categorisations to artefacts of information. Many popular web applications such as Blogging, photo communities and other community-based sharing sites allow users to add tags to their artefacts to give them a more personal meaning that can be shared and used publicly among the community. Tag aggregators at sites like Flickr, Delicious and Technorati show degree of popularity of a given tag via a Tag Cloud which generally uses font size to represent degree of use: more popular tags show up with larger, bolder fonts. Clicking on a tag takes one to a page listing each occurrence of that tag. This result list demonstrates one of the weaknesses of social tagging: these link lists are both undifferentiated and ambiguous. That is, clicking on a tag for a conference may return hundreds of photos in a photo space of that conference, but provide no mechanism for getting at a photo of the speaker of interest. Likewise multiple groups may have multiple meanings for the same tag. There is no way with conventional tags to associate a particular meaning of a tag with its signifier to see only results associated with a given understanding of the term. Because of these problems with number and ambiguity, tags are used mainly to show some sense of levels of interest in a topic within a community, since they are not particularly effective for exploring back into the Tag Cloud to look for interesting content associated with that tag. In this proposal we plan to explore how, by addressing the number and ambiguity problems, social tagging might be redesigned to improve digital repository functionality for cross-repository exploration and discovery. We call this approach Semantic Tagging.  

Aims and objectives 

The aim of Semantic Tagging is to add structure/meaning to tags by connecting keyword tags to relational structures such as folksonomies, taxonomies and/or ontologies. Associating tags with such structures means that tags acquire structured meanings: this tag is related to that tag, which is also related to another tag. With such associated relationships, it becomes possible to infer other, non-explicitly stated relations. For instance, if one is looking at a tag for "mSpace" the associated tags that make up that marker might unpack to include tags for HCI, the Semantic Web, Computer Science, which themselves will unpack to their related meanings. These tags will also be associated with references via Semantic Web technologies to concept definitions, so hovering over "mSpace" for example can expose a definition of "mSpace." By automatically analysing the graph shapes of these semantic tags, relationships with other tag concepts can be inferred. It therefore becomes possible for the system to present not only papers in the repository explicitly about mSpace, but to present research that is in a similar space, such as another semantically aware facetted browser, even though this alternative browser has not been explicitly linked to mSpace. Because of these associated meanings and inferences, the repository becomes not just a place to search for artefacts that explicitly match keyword patterns, a la Google; it becomes a place where it is possible to use semantic tags as tools to build knowledge by being able to explore the context of the artefacts dynamically. Thus, semantic tagging adds new repository functionality by providing tools to assist the research process, such as developing domain knowledge. For example, a visualisation of the tags and sub-tags associated with a concept provides a structured overview of the domain space itself, such as, "this project relates to these three domains, which relate to these particular concepts in those domains." By moving from the semantic tags to the artefacts associated with the tags, one is able to pull up the artefacts via context such as "these are directly associated artefacts; those artefacts are related but not necessarily about the thing in question."

Likewise, domain exploration via semantic tags can be refined to focus such exploration on a given aspect of a tag: if one is only interested in looking at the Semantic Web aspect of mSpace, one can simply turn off the HCI part of the semantic tag space and focus on the area of interest. With such an approach to social tagging, exploration of repositories evolves into knowledge building by supporting domain exploration via explicit and related links of assets across archives.  

Project methodology

Our technology approach, for a technique we are developing called semantic tagging supports the capture, association and sharing of semantically enriched repository artefact categorisation. Our design approach is to use user-centred design and iterative evaluation to assess and refine services supporting Semantic Tagging.

Anticipated outputs and outcomes

• Web services to support the creation of and access to semantic tags associated with IR artefacts for the purpose of cross-repository exploration.
• A test framework to support the publishing and discovery of cross-repository tags.
• Exploration of automatic association of such tags with repository artefacts.
• Evaluations of social semantic tagging in use.
• Exploration of role of Trust in take-up and use of social tagging for EPrints exploration.
• Recommendations both in terms of architecture and usability, stemming from evaluation for the larger scale deployment or adaptation of social semantic tagging.

Lead institution

• Southampton