Remixing Data at ELAG

ELAG  2009 - Metadata highlights, by Janifer Gatenby

This year's ELAG (European Library Automation Group) was held at the culturally and historically important University Library in Bratislava, Slovak Republic, located in the heart of old Bratislava.  The title of ELAG's 33^rd seminar was "New Tools of the Trade" and the conference was full of stimulating and relevant content, with a focus on re-mixing data.

Map metadata is going to get much easier to create and much richer.  Petr Zabicka and Petr Pridal from the Moravian Library in Bruno, Czech Republic,  introducted us to their web site oldmapsonline where there are open source tools for the scanning, metadata creation and adding geo references to maps.  Their open source Map tiler provides an easy interface for assigning a geo bounding box that makes the map compatible with Google maps so that maps can be overlaid, e.g. for "then and now" comparisons. Geo coordinates are more important than traditional access points for searching maps.  They also recommend  zoomify which zooms any image.

To open the main theme of the conference, Karen Coombs gave a rich key note address and animated the mash up work shop. Check her presentation and her workshop notes, both of which are full of useful examples and tips.

Table of content metadata is being harvested and made available for reuse.  This was reported by Lisa Rogers from Heriot-Watt University in the UK with her overview of TicTocs and Golddust. TicTocs aggregates RSS feeds from more than 12,000 journals and then makes a data set available for mash up.  Peter Van Bohemen from Wageningen University has made very rapid use of this service to display the contents of the lastest issue of a journal when a full record display of a serial is requested in the Wageningen union catalogue.  Gold dust is an SDI service using Tictocs and user profiles. 

There were two reports on systems with a new approach to the generation of recommender data.  Marcus Spiering of the University of Karlsruhe reported on Bibtip, which is a recommender system based on evidence from an anonymous session based cookie that looks for "co-inspections" (full record views).  This metadata is harvested from the usage information collected from a library's online catalogue and thus it works for all material represented in the catalogue.  This contrasts with recommender systems based on circulation based usage which only look at the physical collection and systems based on resolver usage which only look at electronic material.  Tamar Sadeh from ExLibris announced bX, a journal article recommender system based on traffic from harvested logs from SFX resolvers. It also looks for "co-inspections" within a session and is based on research from Herbert van de Sompel's Los Alamos lab.  ExLibris will be running this as a chargeable web service.

Thom Hickey and I gave a presentation entitled "Opening Library Data for Web Scale and Re-mixing" . Tom talked about our data resources and how OCLC is both growing and enriching them, with examples from WorldCat Identities and VIAF.  I stressed the importance of identifiers in re-mixing data, alluding to GLIMIR (Global Library Manifestation Identifier:  see my post in January 2009 on the importance of identifiers) and presented an outline of OCLC's identifier services and data APIs.  From the discussion that ensued, we gathered that work identifier services are in demand. Increasingly, metadata specialists are recognizing the importance of manifestation level identifiers as well.  See, for example, the post this week by Jonathan Rochkind.