Last Frday, 3rd, September,  the Metadata Forum held a second face-to-face meeting at the RepoFringe 2010. With an event focussing on repositories, there was much talk of metadata in both the practical and the abstract.  Talks and sessions from the RepoFringe can be found at the event blog. The blog is being added to over the next few days, so please check back if a specific session is not yet ‘live’ on the blog.

Tony Hirst of the Open University gave an interesting and thought-provoking keynote address, looking at mashups and using informal publication methods. He raised the inevitable questions around the default use of the PDF format for documents, and it’s many limitations with regard to accessing content. This was only matched later in the evening by his insightful thoughts on popular music and how to freak out your teenage offspring shared at a nearby alternative Fringe venue.

Both days saw Pecha Kucha sessions, with speakers being given 2 minutes and 20 slides to share their projects. All the speakers were excellent, though explaining an institutional repository through the medium of cake seemed a sure winner for the first day for Robbie and Toby talking about Enlighten, the IR of the University of Glasgow! Do check out the session slides as they become available – they are an very useful way of getting a quick overview of many exciting projects happening at the moment.

Round table sessions on both days provided a good way to discuss topics of interest in more detail. The Metadata Forum chaired one of the sessions on… metadata. Specifically we looked at metadata for time-based objects and the aggregation of metadata, helped by Sheila Fraser of EDINA who is working on a JISC Scoping Study on this subject. Full details will follow in my next post.

Other particularly interesting presentations from a metadata perspective were given by Herbert Van De Sompel of Los Alamos National Library, Chris Awre of the University of Hull and Michael Fourman of the University of Edinburgh.

Herbert looked at adding a time dimension to searching, allowing people to search for earlier versions of web pages through the Memento project. The implications of this were very interesting and the idea of being able to search in time as well as space was intriguing and useful.

Chris spoke about the Hydra project, a joint project between the Universities of Hull, Stanford and Virginia working in partnership with Fedora Commons. The success of this collaborative venture, where each institution offers it’s expertise and resources to create a repository that can be tailored to individual requirements yet would be out of reach without thus collective approach was inspiring. Agreements and flexibility is the way forward – something that affects metadata decisions on a very practical level.

Michael explored a different approach to classifying documents using Topic Models. Topic Models offer an alternative, potentially more accurate way of classifying documents than keywords, as they pinpoint more accurately what an article is actually about by looking at word frequency. A little Bayesian magic is then applied to balance the model.

The closing address was delivered by Kevin Ashley of the DCC who gave a great summary of the past two days. He also spoke about the need for repositories to work with and link into many other systems – a vision that is, of course, supported by metadata in many forms! Kevin’s humorous yet thoughtful address made a perfect end two an event that had been both useful and enjoyable.

Thanks to all the organisers for a great RepoFringe 2010. The Metadata Forum would also like to thank James Toon of the ERIS project for sustaining the metadata frenzy of the Fringe with possibly the best Chinese dumplings she has ever had the pleasure of eating.