Why bother with linked data?
In my last blog post I gave a brief introduction to the linked data offered by Ordnance Survey. Furthermore, in that post I said ‘By linking to identifiers for places and postcodes in your data you can enrich the information you hold’. It is this point I want to expand on here.
Last year Talis, Iconomical and the data.gov.uk team got together to build a Research Funding Explorer. The idea behind this (as explained here) was to show how linked data techniques could bring real benefit when it comes to joining and analysing data from a number of different sources.
The original application allows a user to browse funded research projects by subject and organisation. Furthermore, the original work contained a geographical element which allowed you to see which projects were being funded in which region.
Linked Data
Following this link we can see which projects are being worked on by institutions in the South East of England, and how much funding they are receiving. However, it would arguably be useful if you could also view the geographic distribution of projects at a more fine grained level. This can be made possible by linking the data to Ordnance Survey linked data.
Here’s how…
In the original data the postcode is recorded for each institution. The postcode is captured simply as text. However, replacing the postcode text with a postcode URI means your data suddenly has the potential to contain a lot more information. A simple step in a text editor will take you from the text “SO16 4GU” to the URI data.ordnancesurvey.co.uk/id/postcodeunit/SO164GU.
In doing so we can now ingest the linked data for that postcode.
Because the postcode linked data provides a look up between postcode and region, we can enrich our original data with knowledge about the ward, district and county (where applicable) for each institution. This means we can now analyse research funding by local authority area as well as European region.
Using the spatial relationships in the Ordnance Survey linked data means we can also start to do more complex analysis. For example, we could compare funding in one region with funding in its neighbouring region or use the containment relationships to aggregrate the information up to coarser grained geographies.
None of this could have been done with the original data, but is made possible simply by replacing some text with a URI.
Hopefully this simple example shows that using linked data techniques to bring two relatively simple datasets together can start to yield benefits. The braver readers amongst you can read about this in a bit more detail on my personal blog.
Photo by Reedster via Flickr.
Themocracy
A post code on it’s own can be re-used by anyone able to, an OS URI ties it to OS and that particular tool….
I don’t think it’s good to encourage people to alter generic data to system specific data, though of course it looks good for OS.
Including a URI like that in ADDITION to the plain postcode would at least provide the functionality without hindering potential resuse of the postcode data for other purposes.
Adrian – yes it depends on your application. You can either change the postcode to the URI if that suits your needs, or in some cases it might be more appropriate to add the URI in addition. The point was more that in a linked data world a postcode URI can be a lot more useful than a textual representation of that postcode.
However, what do you mean by ‘generic’ data?
Well a plain postcode can be used for whatever variety of purposes someone can come up with, used with any tool that handles postcode data. A postcode obfuscated by a URI specific to a particular tool restricts it somewhat. Anyone able to access the data and wanting to use the postcode for something else would first have to clean it up.
Generally I’d expect a postcode field to be left as such and to build the URI on to it, considering the standard format.
I’m actually in the process of cleaning up the underlying data, and, yes, /adding/ in the post code links. Hope to have something to say about this in public within a week. It’s a long job as there’s quite a bit of manual labour involved in cleaning up the data, spotting errors, etc. I’m using Dave Reynolds’ Org vocabulary, which entails using VCard as well
.
John,
As you point out, the potential is massive, and its exactly the reason I and many others have been campaigning for access to OS and other government data for years. For this to be a success, you have to be sure that URI will still be around in 1 month, year, decade. Now given the OS has resisted the opening up of her data chest, and is still led by a CEO who doesn’t believe in open data, and is on record as having blocked the opening up of OS data on multiple occasions, how can we expect this linking of data to deliver on its potential?
I would love to use data in this way, but I’m not alone in my reservations of the commitment of the OS to open data.
Jack
Hi Jack,
This is probably a silly question, but have you heard about OS OpenData? It’s provided free access to Ordnance Survey data for commercial and non-commercial use since April last year. http://www.ordnancesurvey.co.uk/oswebsite/opendata/
Paul
Jack,
In addition to Paul’s link, this is probably worth a read:
http://blog.ordnancesurvey.co.uk/2011/01/changes-to-the-os-opendata-licence/
Meanwhile, “Get Excited and Make Things!!”
John
[...] this blog post I talked about the potential of (Ordnance Survey) linked data. Partly motivated by this challenge I [...]
[...] the driver behind the Ordnance Survey’s publication of some of its data, has written several blog posts recently that give examples of how the OS post code data can be linked. In his posts he kindly [...]
[...] the driver behind the Ordnance Survey’s publication of some of its data, has written several blog posts recently that give examples of how the OS post code data can be linked. In his posts he kindly [...]