Linked Data – turning the web into a database

all aboard the semantic web

Last week I went to an event that looked in some detail at the Talis Platform, now emerging after 5 years of development as a semantic web environment that offers a hosting and development infrastructure for data owners and publishers, and for developers who want to build applications using that data.

The Platform is designed to encourage sharing of the linked data that will underpin the next generation of web applications and data services. To paraphrase Ian Davis, CTO at Talis, currently the web is a universal information space for documents; linked data will make the web into a database.

Last week the UK Government launched the open data project, which seeks to open up the vast quantities of non-private public-sector data stored on government computers. The idea is to encourage innovative re-use of the data; there are nearly 3000 data-sets available for developers to build mashups with and interesting applications are already starting to  appear; examples mentioned in a recent BBC blog include a school finder which lets you search local schools ranked by Oftsed score, and FillThatHole, which uses ONS Census geography data to facilitate the reporting of potholes and other road hazards. The Talis Platform is providing the hosting and search services behind the site.

Library & Learning Resources are also using the Platform, in our case to deliver the latest incarnation of the library catalogue.

Prism 3 will be hosted by Talis and delivered to us using the Software as a Service model; because the data and the application are both on the Talis Platform, rather than on servers here at our own data centre this makes Prism 3 an example of what’s often called cloud computing.

Our bibliographic data has been copied into a self contained data store within the Platform and converted from MARC into RDF, which is the data format for linked data (if your inner geek wants to find out more about RDF then here’s a place to start). By utilising data held in other data stores within the platform Prism 3 will be able to deliver enrichments such as images of book jackets to give the catalogue a lot more visual appeal for users.

There should be a basic version of our Prism 3 up soon for you to play with, so please watch this space.

image credit: Thomas Lennon Photographic Collection, Powerhouse Museum

So close, yet so long.

Two seemingly straightforward requests received in as many days, produced groans when I first knew about them. Now, I’m not starting a post Christmas slump, one from which I would be unlikely to emerge until the clocks goI wouldn't fancy dropping this on my foot forward at the end of March.  The requests themselves don’t require a feat of mental gymnastics such as that needed to consolidate the different spellings or misspellings that we have for Russian authors in our catalogue. So where’s the problem.

The requests were: how many items were returned yesterday and what transactions were processed at a Campus Library around a particular time on an afternoon in December.  Straightforward certainly, however frustration arises, because for these requests it takes so long to get the data out of the Library Management System. The report ends up grinding through nearly 21 million loans; and can’t be run on our live system for fear of bringing it to a halt.

I’ve worked in Libraries for the past 5½ years and before that I worked for a small company producing Business software.  Even back in the dim distant past of the early 90’s, up to the minute sales and transaction reports could be produced by end-users without any impact on the live system. The slowness of the current reporting is due to database design not just system speed. I know, Libraries have very different demands and pressures from commercial companies, but it is still galling to know that I could have produced equivalent figures in a couple of minutes, when mobile phones were still the size of a brick (or two).