Scottish Linked Data hackday report
We had another successful Scottish Linked Data hackday last week. Around a dozen keen linked data folks spent a day at Inspace, part of the Edinburgh University Informatics Forum, working away on a variety of linked data projects. Several more folks joined for the end of day presentations and discussion session.
A brief summary of the end of day session:
Peter Winstanley presented his work on use of Semantic MediaWiki for creating a knowledge base around who works on what at the Scottish Government, and how to make it easy for regular users to add meaningful markup.
Keith Alexander continued the work started at the previous SLD hackday, on creating a linked data version of the Scottish Neighbourhood Statistics data.
With the hackday taking place just after the Wimbledon finals, Alejandro Mallea’s topical project was to create an RDF version of data on results of ATP tour tennis matches. He was developing a vocabulary for his data model, based on an extension for the BBC sport ontology. He loaded the RDF results into a Virtuoso triple store and demonstrated the kinds of detailed analysis enabled by SPARQL queries on that store.
I talked about making Linked Data for ALISS, together with Derek Hoy, who with Andy Hyde has been responsible for building the ALISS system. ALISS aggregates information on resources available to people with long term health conditions and makes those available via a website and API. We are working towards adding a Linked Data API and SPARQL endpoint as an additional way of accessing this data.
Troy Giunipero was working on visualisations of statistics of links between different linked datasets, as a way of asssessing alternative approaches for deciding whether two URIs represent the same resource – in order to automatically generate links between different datasets. He was using tools from the Silk project (“A Link Discovery Framework for the Web of Data” from the Freie Universität Berlin).
We had an excellent discussion around the presentations and related topics (thanks particularly to Rob Stewart for his many probing questions!) The problems of stale linked data and how best to avoid it was discussed at some length, leading into discussions of versioning, provenance and trust and alternative approaches for representing that kind of metadata in RDF. William Waites is currently collecting examples of stale linked data with the aim of carrying out some research on the topic, so let him know if you have any good examples!
Many thanks to Ewan Klein of Edinburgh University for arranging the excellent venue and copious supplies of coffee, buns and pizza. Thanks to all who attended and contributed to an interesting and useful. Discussion on a variety of semantic topics continued in the Pear Tree afterwards and seemed to be still going strong when I had to leave around 7.30.