ISKO-UK Conference on Linked Data
September 15, 2010 1 Comment
Yesterday I attended the ISKO-UK one-day conference on Linked Data. I have to admit to attending with an interest (what will be the impact of Linked Data on institutional repositories?) but also a degree of scepticism. The Semantic Web/Linked Data movement is about ten years old now – ten years that I spent in commercial web companies with some high profile clients but never once had a query regarding Linked Data. We all know about the growth of the web in its first ten years but Linked Data seems to be finding it harder to catch on. Perhaps this is because there doesn’t yet seem to be a compelling commercial application. It was telling that the commercial companies on the attendees list were generally Linked Data service providers, not commercial companies looking to find out what Linked Data can do for their company. I think one of the main drivers of growth for the web was that companies realised that they could use it to make money and it isn’t clear, to me at least, how this applies to Linked Data at the moment.
That said, I don’t want to sound like I’m putting the conference down because it was a very interesting day densely packed with informative talks. One of my main aims in attending, with my sceptic’s hat on, was to try to find out what real applications are being developed using Linked Data. In my view the most interesting talk, because it does have a real world commercial application, was Martin Hepp’s overview of GoodRelations, a standardised ontology for publishing product data. They are initially focussing on embedding RDF-A and this has already been adopted by Yahoo to enhance search results (Google and Bing may follow soon). However, the potential is there to provide SPARQL end points that could allow for much more accurate aggregation of product data for a new generation of shopping and price comparison sites.
John Goodwin also provided an excellent overview of the Linked Data work being done at the Ordnance Survey. This was a good demonstration of the potential of Linked Data to provide applications drawing in information for multiple sources, going somewhat further than the ‘standard’ geographic mashup demonstration overlaying a single data source on a map.
As would be expected, Nigel Shadbolt spoke eloquently in the morning keynote about government open data initiatives and this seems to be something that the Linked Data community is pinning its hopes on. The importance of this shouldn’t be underestimated – I believe that opening up government data to scrutiny and mashups with other data sources has the potential to genuinely help to hold governments to account – but again it isn’t a commercial application. If the community focusses too much on this area is there a danger of Linked Data becoming a public sector concern that doesn’t have much impact on the web at large? Is this a big issue in the UK and US because we have governments that are open to the idea of publishing data and heavyweights like Tim Berners-Lee pushing the idea to them? As Bernard Vatant said during his presentation, lamenting the lack of open government data in France, “We don’t have a Tim Berners-Lee”.
An interesting undercurrent in the conference was a certain shying away from the development of ontologies and OWL. This recent blog post from Chris Gutteridge was much discussed and there seemed to be implicit agreement with his statement that “OWL appears to be the hardest part of the whole RDF thing”. A more lightweight approach was suggested using SKOS, which was summarised by Antoine Isaac. In presenting the PoolParty application, Andreas Blumauer suggested that SKOS was adequate for modelling in “most” scenarios. Perhaps this route may help to reduce the perceived barriers to using Linked Data.
I realise that this post has sounded rather negative, which doesn’t give an accurate representation of my attitude following this conference. I am enthused to continue to develop an interest in Semantic Web and Linked Data technologies. The conference didn’t dispel my scepticism entirely but I’m sure Linked Data wouldn’t be interesting if there weren’t challenges ahead.