Wednesday, 26 June 2013

LODLAM 2013

I attended the LODLAM conference in Montreal last week. This was the second Linked Open Data in Libraries, Archives & Museums conference and around 100 delegates attended the National Library and Archives of Quebec from cultural institutions, universities, private companies such as Axiell and other organisations from around the globe such as OCLC and the BBC. Attendees included representatives from the US, Canada, UK and Europe, New Zealand and Australia, and other countries.



National Library & Archives of Quebec

The un-conference format provided plenty of room for brainstorming and breakout sessions on a multiplicity of themes - most running in parallel, which meant there wa a lot of running between seminar rooms! The conference also had a competitive element in its run-up, where teams of Linked Data experts developed or refined showcase projects. The semi-finalists lined up before four X-Factor judges before a winner was announced. Networking was also the name of the game as we tried to link up our respective projects and think about the development of more robust services - many of us ran out of business cards, I am sure.

The breakout sessions covered all areas of Linked Data, not least controlled vocabularies, mapping and patterns, the user experience, teaching Linked Data, ontologies such as CIDOC-CRM, natural language processing, historical mapping, annotation tools, World War One and many other subjects. Several sessions were an opportunity to view new tools such as KARMA, GNOSS and PUND.IT, which have been designed to facilitate the mark-up and annotation of web pages and the aggregation of different entities and types of information. One very informative session covered the Getty release of its vocabularies as Linked Data, which should prove enormously useful to the whole community. The World War One session was an opportunity for participants to gain an overview from across the globe, including trench mappers and authority experts from the US, Finland, Australia and New Zealand. The BBC showed off its World Service audio file transcription project which contains a significant crowdsourcing element to help refine data.

The LODLAM challenge finalists were universally excellent. Linked Jazz showed off its superb visualisation tools to depict relationships and influences between jazz artists. Free your Metadata was a song combo that got the floor buzzing with its appeal for data reconciliation and cleansing as a prerequisite for successful linking. Mismuseos combined museum metadata and images for museums. The winner was PUND.IT, a... 'client-server annotation system which lets you express semantics about any kind of web content through labeled relations among annotated items, linking them to the Web of Data. Annotations can be shared and organized into private or public notebooks which can openly accessed to build engaging visualizations'.

The conference as a whole showed the vast array of work that has been undertaken throughout the GLAM sector to create new datasets and develop or refine tools that will allow new, more robust, services to be developed - a good mix of the theoretical and the practical. The need for training that demystifies the subject, a stress on users including the front end experience and visualisation, and quantifying the benefits to organisations through use/economic cases appear to be priorities. 


Montreal

Monday, 17 June 2013

Historypin mapping tool

Good news that Historypin have completed development work on a test version of the new mapping tool for displaying Linked Data links from archive catalogues on the Historypin Google maps interface.

The tool interrogates a portion of the many thousands of place names associated with catalogue entries on AIM25, which aggregates collection level catalogue descriptions from archives held in around 130 institutions in the London area, including learned societies, universities, museums and local authorities. AIM25's 17,000 catalogues relate to collections containing a wealth of information about every corner of the globe, spanning around 500 years of history and covering topics as diverse of scientific discovery, war and international relations, exploration and travel, biography, politics and religion.



The mapping tool was an experimental component of Step change, which sought to release the UK Archival Thesaurus (UKAT), the key UK subject vocabulary, as a Linked Data Service; to develop a new editing tool for the creation of semantic archive catalogues; and  to embed the tool in Axiell's CALM, one of the most popular proprietary archive cataloguing applications used by some 400 institutional customers in the UK and Europe.

The mapping project interrogates UKAT for relevant 'Scope and Content' related place names (ie catalogues tagged up with place names relevant to the source material and not biographical or other contextual information). These catalogue titles are then flagged on the map, which is badged as a Historypin themed channel. Users can expand titles to read relevant information about the collections and make arrangements for visiting the archives or to follow links to digital surrogates of the source material.

The sub-project was not without significant obstacles. Issues included performance associated with too many simultaneous queries; difficulty in visualising different levels of granularity - what constitutes a 'place' - geo-co-ordinates, a bounded area or a subjective understanding of the 'local'; and the problem of flagging up diverse locations listed in single entries (something that often happens with personal paper collections of individuals whose lifetime careers might include study, emplyment or family life across the world). Work arounds included the inclusion of a granularity filter to overcome the problem of excessive clustering of returns; and limits placed on simultaneous queries. The resulting map only returns a portion of the many tens of thousands of places described in AIM25 and doesn't yet link with a conventional Historypin map showing pinned photographs of local places. This will follow soon, to develop a more useful and integrated service combining photographs contributed by partner archives, along with relevant catalogue entries that provide useful contextual information.