About the author
Dan Rickman has over 20 years experience of the geospatial industry working as a strategist, project manager, technical consultant, business analyst, systems analyst and developer in public and private sectors
We had an excellent turn out last Thursday for our first afternoon seminar (around 45-50 people) and a set of complementary and fascinating presentations.
We'll make the slides available on our web site or via slide share as soon as possible. Conrad Taylor has made a recording of the afternoon and we plan to make this available soon – we are considering how to link this to the slides. Our thanks go to Conrad and a number of people who helped with the afternoon, not least all the speakers and panel members and also members of BCS KIDDM including John Lindsay from Kingston University who also brought along some fascinating exhibits.
I gave a general introduction to the day based around the idea that information is classically defined as data + context, and context is now provided in a sense by "semantics" in the form of RDF as a pragmatic approach. This is being driven in practice through the work of Tim Berners-Lee and Nigel Shadbolt (a former BCS President) on the linked data initiative and data.gov.uk
In his Semantic Web Roadmap, Tim Berners-Lee describes the semantic web as turning the World Wide Web into "... a set of connected applications ... forming a consistent logical web of data..." (Tim Berners-Lee, Semantic Web Roadmap). We had a previous presentation [pdf] related to the geospatial web from Jonathan Lowe in which he described the semantic web as being able to make "Rain Man" style connections. Tim Berners-Lee provides an example where two databases, constructed independently and then put on the web have been linked by semantic links which allow queries on one to be converted into queries on another. His specific example uses location - one can link information about employees and their location to one about friends and their location thereby allowing one to identify people who are friends of employees – of course, whether this is desirable raises a set of issues. It is clear that spatial data (in fact spatio-temporal data) is key to the development of linked data.
John Goodwin from Ordnance Survey then gave a very clear description of the Linked data initiative and how it works through RDF "triples" which define subject, object and relationship between them where the subject is a URI or an XML literal. Combined with the fact that a URI provides a response dependent on whether it is called from a web browser (when it returns a web page) or an automated link (when it will return RDF), this provides a core infrastructure for linking disparate web-based resources on a common basis. These can be navigated using services such as sameas.org NB: see "How to Publish Linked Data on the Web" for an overview (NB: not mentioned in John's talk). He then provided some great examples and stressed the centrality of nodes in the linked web such as dbPedia and Geonames
Permanency: place whether existent or historical never stops being subject of documents(One studies and writes about Roman Empire, Habsburg Monarchy or USSR long after these cease to exist)
Parallelism: contemporary administrative subdivisions have to exist in parallel with traditional regional subdivisions (e.g. Dalmatia or Slavonia are not administrative units in Croatia, Povardarieis not an administrative unit in Macedonia )
Associative linking: contemporary geo-political subdivisions are linked to related historical entities, traditional subdivisions and physical geography
Classifications seem to be well-equipped for:
I would add that the Linked Data world is very much looking to re-use existing classifications and place name (specifically currently Geonames) is becoming a key hub in the emerging world of linked data. The "Simple Knowledge Organisation System" is one of the key elements being utilised to create semantic links via RDF
Christopher Osborne from ITO World then added a complete contrast in a presentation which questioned the fundamentals as to why data.gov.uk is a Good Thing. He argued passionately that this was a requirement for an open democracy and would lead to a process of conversation with citizens rather than the current process of “consultation”. He therefore stressed that for him technology was not of concern – what was required now was commitment to a more open process. He urged that government departments should not be allowed to get away with delaying provision of data because they did not have skills or resources to put it in RDF format, any format will do at least for initial release and to get the process established.
We were also very fortunate that James Forrester from the Digital Engagement team of the Cabinet Office was able to join the panel discussion at short notice. James explained that he is half of this team and that they were working closely with government departments to help them publish their data via data.gov.uk
The panel discussion was quite wide ranging and there were many more issues which could have been discussed. It was suggested by people in the audience that this is the most significant change in government data policy that we have seen and it represents a major cultural change for the public sector. There were nevertheless concerns from some public sector representatives there regarding funding models – data may be free at the point of access however it is far from free to create and maintain. In addition, some organisations were concerned with the principle of private sector organisations creating revenue streams from their data.
From a technical viewpoint, the view was expressed on the panel that RDF provided significant benefits over relational databases and that this represented a paradigm shift in the same that the relational databases themselves replaced hierarchical databases [check reference]. If so this raises a host of issues - RDF here becomes not just the “context” which turns data into information but also the data itself. Whether this is necessary, technically correct and commercially viable given the investments already made to date in structured relational data remains to be seen but appears unlikely and not the intention of the designers of the semantic web.
The conversation could have continued well beyond the planned finish time for the event. You can use this blog to continue the discussion and I have asked the speakers to participate as well – here are some issues for a start:
I look forward to continuing the discussion.