Reflections on parallel sessions: Dynamic tagging and wisdom of the crowd (Durham Blackboard Users Conference 2011)

This post is a summary of thoughts of one of the parallel sessions at the 2011 Durham Blackboard Users Conference. The session presented a way of sorting a list of resources based on location, using social tagging to rank and categorise various resources.

Session: 'Dynamic Content Ranking using Location-based Folksonomy' by Dr Mike Batty, University of Durham.

This I found to be a fascinating and enjoyable trip back to my engineering degree as this session included some tasty mathematics and computer coding wizardry. Essentially, the ‘Technology Enhanced Campus’ project uses the wi-fi hotspots around campus to triangulate the location of a mobile device. On this mobile device a student would add tags about what is relevant in that area (e.g. café, department name, social activities). When the tags are used by the student, wandering from location to location, they grow/diminish in size relative to the original tagging location – this is the dynamic ranking effect.

Aside from this, what was proven by Batty’s work is the ‘wisdom of the crowd’. Using search engine precision and recall graphs, the relevancy of tags created by ‘the crowd’ (all users tags collected together), as opposed to the tags created by the individual, was greater. Therefore, the contributions of ‘the crowd’ had a positive effect by adding new meaning and context to the student.

Taking this out of the ‘location’ theme and applying to learning resources, we can surmise that if a student tags resources they are interested in, they may only be able to find a limited range of other resources they may also be interested in. However, if they select resources which have been tagged by multiple people and use the collective tags to find new resources, chances are that the new resources will be just as relevant, if not more so than if the student had only used their own tags.

Interesting stuff and I see a lot of potential in both harnessing more crowd-sourcing activities and dynamic ranking. Perhaps it is possible to use other proximities (as in John Traxler's keynote) such as time, history of previously used resources, etc as a way of dynamically ranking tag clouds. Might take this up as a project.