Category: Semantic Tech & Business 2011

A guest post by Neil Wilson, Head of Metadata Services, British Library

I have been doing a number of presentations recently concerning the British Library’s new range of free data services and particularly the ‘Linked Open BNB’ that we launched via a Talis platform in July which as just appeared on the latest LOD Cloud diagram. Most recent of these was one for the Semantic Technology & Business Conference (London : Sept 26/27). However, before I come to some of the points I was trying to get across; a few words about the conference…

One of the objectives of the BL’s new open metadata strategy is to engage with the wider community in order to try and move beyond the traditional ‘library silo’. Semtech certainly provided an opportunity to do this since it was chiefly attended by a wide range of business technology implementers in addition to public sector organisations. Most interesting to me however was the fact that such a wide cross section of organisations: from the US Dept. of Defence, to the Amsterdam Fire Service by way of the BBC and a variety of hard nosed commercial companies were not just experimenting with triple stores etc or building them into future plans but actually using them in critical applications right now. It was therefore highly encouraging to see that the possibilities and value presented by semantic technologies were being made real by such a diverse variety of companies and public sector institutions. Undoubtedly there is still a way to go until the Semantic Web becomes a reality but the fact that applications are moving from the experimental to the everyday should help to accelerate its development further and convince others to take the plunge.

Back to my own presentation:

I really wanted to address some queries I had received from other organisations covering three main areas:
• Why is the BL experimenting with linked open data?
• What choices might you encounter when creating a linked data service?
• What lessons have you learned?

I was interested to see that some of the latter (e.g. develop incrementally & ‘grow as you go’ or exploit existing tools, standards & expertise to get an early result) were echoed in several other presentations. From some of the feedback received after the talk, at least some of the points made may have helped those contemplating a similar ‘linked data journey’.

So has it all been worth it? Statistics (i.e. 250K transactions in Month 1) suggest the BNB service is certainly being used. While coming to terms with linked data was a steep learning curve, colleagues found the project a highly positive experience. The BL also benefited from being one of the first library sector movers in the area; with the system assisting both visibility and relevance while also suggesting new options for future development. In a wider context, I would argue the results also show: libraries can rise to the challenges of a rapidly changing environment; have valuable resources to offer and an important contribution to make.

This post was initially just going to be about the presentation The Simple Power of the Link that I gave in the opening session of The Semantic Tech & Business Conference in London earlier this week.  However I realise now that it’s title, chosen to draw attention to the core utility and power of the basic links in Linked Data, has resonance and relevance for the conference as a whole.

This was the first conference in the long running Semtech series  to venture in to Europe as well as include the word business in it’s name.  This obviously recognises the move, from San Francisco based geekdom to global pragmatic usefulness, that the Semantic Web in general and Linked Data in particular is in the process of undertaking.  A maturing of the market that we in Talis Consulting can attest to having assisted many organisations with their understanding and adoption of Linked Data.   In addition to those attendees, that I would characterise as the usual suspects, the de-geeking of the topic attracted many who would not previously of visited such a conference.   An unscientific show of hands, prior to the session in which I presented, indicated that about half of the audience were new to this Semantic stuff.

Traffic on our stand in the exhibition area also supported this view, with most discussions being about how the techniques and technologies could be applied, and how Talis could help, as against esoteric details of the technologies themselves.  Linking people with real world issues and opportunities, with the people that have the experience to help them, was an obvious benefit of the event.  In addition to as having some great keynotes, as Rob described.

So, back to my initial purpose.

The presentation “The Simple Power of the Link” was an attempt to simplify the core principles of  Linked Data so as to highlight their implicit benefits.  The aforementioned geekdom that has surrounded the Semantic Web has, unfairly in my mind, gained the topic a reputation for being complex and difficult and hence not really applicable in the mainstream.  A short Google on the topic will rapidly turn up a set of esoteric research papers and discussions around things such as inferencing, content-negotiation and Sparql – a great way to put off those looking for an easy way in.

There is a great similarity to the way something like vehicle engineering is promoted with references to turbos, self levelling suspensions, flappy-paddle gearshifts, iPod docks and the like – missing the point for the [just landed on the planet] new to the topic – that the major utility of any vehicle is that it goes, stops, steers, and gets you from A to B.

The core utility of Linked Data is The Link.  A simple way to indicate the relationship between one thing and another thing.

As things in Linked Data are represented by http URIs, which when looked up should return you some data containing links to other things, an implicit web of relationships emerges that you can follow to obtain more related information. This basic, and powerful, utility could be simply realised with data encoded as triples in RDF, served from a simple file structure by a web server.

So, although things like triple stores, OWL, relational-to-RDF mapping tools, named graphs, SPARQL 1.1, and [the dreaded] httpRange-14 are important issues for those embedded in Linked Data, the overwhelming benefits that accrue from applying Linked Data come from those basic triples – the links.  As a community I believe that we can be rightly accused of not making that clear enough.  Something that my colleagues in Talis Consulting and I attempt to address whenever possible.  Especially at our open Linked Data events.

We knew that Semantic Tech & Business in London this week was going to be a great conference with some real business message but we couldn’t have predicted how excellent the keynotes were going to be.

Straight from the recent announcement that Volkswagen are using extensive semantics for their product data we have Martin Hepp presenting the way that structured data enhances the web. Martin gave great and essential messages, describing how rich product data is destroyed by the web today. He describes the web of documents (quite rightly) as a data shredder.

Through several major points the other that hit me between the eyes is how so much effort is spent optimising the experience of a web page once a visitor has landed there — yet the web has evolved (and is evolving) to show users key information without visiting the page. That means we have to invest far more in optimising for the way your data displays before a user arrives. Richard has been blogging about the use of Linked Data and Semantics in SEO and SERP for a little while now and if you want to discuss how to make the data on your site work harder to get visitors to come to you then we’d like to talk

Steve Harris of Garlik talked about the way they’ve used semantic technologies internally at Garlik. Their customers and partners, on the whole, don’t know that they use technologies like this — they’re just impressed by what Garlik can do with the data. He raised some great points, hiring expertise in this area is hard, so they look for good software engineers and then train them in Linked Data and SPARQL. Their experience, like ours, is that developers who have built systems this way for a few months do not want to go back to SQL.

If you have a team of software engineers, developers, data owners, DBAs and project managers who you want to understand this technology then we have a proven two-day training course that teaches Linked Data from the basics.

Steve’s other key message is that this stuff is ready and possible for companies and it has allowed Garlik to do stuff they couldn’t have done with relational technologies.

John O’Donovan entertained us with a seemingly endless stream of the most wonderful (badly phrased) headlines. For him these demonstrate the need for comprehensive and well-managed metadata. He talked about the BBC’s World Cup 2010 project which built its site atop a triple store. Talis Consulting have trained many of the developers and information architects at BBC in semantic technologies.

John mirrors the message from Martin and Steve that this technology is ready, capable of delivering large production systems and has real benefits in terms of power, flexibility and cutting implementation costs.

We’ve been seeing this market mature year on year for some time now and it’s great to see three high profile keynotes all saying the same thing — Semantic Technologies are ready for you to use.

If you want to start using them, come and chat with us