This talk by Richard Wallis from Talis. Talis are basing their approach on the Talis ‘Platform’ which they describe as a Semantic platform – this talk is meant (I think) to say something about what they mean when they talk about a semantic platform.

Richard starts by covering the development of the ‘library catalogue’ – from hand written cards, to the latest iteration of the Talis online catalogue. Now covering the different approaches to ‘union’ interfaces – z39.50 vs physical union of records.

Now, demonstrating problems with searching for ‘paris hilton’ when you want the hotel – if you use Trip Advisor it’s great, because the context is explicit, but if you use Google, then you get a lot on the person. He does another example with Ford Prefect using Google and ebay. To some extent this is true, but in both cases Google managed to return Hotel, and the car respectively in second place on the search – so actually this suggests that Google does a pretty good job. I’ not necessarily saying you get ‘better results’ from Google – but it shows that Google does a pretty good job – although it doesn’t know what you want, where there are two meanings the 1st and 2nd hit in these scenarios exemplify the two meanings.

Richard is just playing the video of the sketch about the first IT professional – to everyones amusement.

Anyway, Richard is arguing that Trip Advisor and eBay work better for the examples above because they have good metadata. I’m not sure about this – they have context as well, so at least it isn’t just about the metadata – if Trip Advisor ‘catalogued’ celebrities as well as hotels, then would it have been any better than Google? I’d guess not on a ‘all fields’ search. To take a slightly contrary example, if you search for ‘Ford Prefect’ on Amazon then you might well be looking for the Hitchhikers Guide books – but what you get is manuals for the car.

So – Richard’s point is that libraries have standardised metadata – so we should be able to exploit this.

Moving onto a different tack, Richard is describing the drop in cost of both storage and computing power. You can now buy a laptop for under £250 (http://www.asuslaptop.co.uk/products.php?cat=53)  (I so want one)

So – the Talis Platform – big data store – about bringing data into a single store – but more than that. However, difficult to describe as it has no user interface – Richard says it’s like trying to describe a new petrol – he can say it’s better, but how to show it? Talis have now started to build interfaces on top of the Platform (for those who are interested, it’s an RDF store) – I’ve seen a few demos yesterday and today of products built on the platform, and there is an online demo of their ‘Engage’ product built on the platform – this is for community information. The point Richard makes is most of the power of the product comes from the Platform – the interface is quite a thin layer over the top…

So – starting to talk about the ‘semantic web’ – what are semantics?

Semantics (Greek sēmantikós, giving signs, significant, seebma symptomatic meaning, from sema (σήμα), sign) refers to aspects of meaning, as expressed in language (from Wikipedia)

The Semantic web is about being able to express meaning relating to content in a machine-readable way, so software can start to link content together based on meaning. At the moment there is some semantic meaning in links – and this is one of the things that Google exploits – the wording I use in the text of the link gives some meaning to what I link to (which is why I try to avoid using links like this)

The Platform is structured in a way that can start to exploit the semantics implicit in metadata – and obviously specifically library metadata (although not exclusively, as the Engage product show, you can apply it to other non-library metadata just as easily)

Once you have metadata in a semantic format, then you can start layering different interfaces on top. If you have a standardise the representation of the metadata – then anyone can layer tools over the top (an example is the Relation Browser)

I think what I need is an introduction to some of the ‘under the bonnet’ stuff – I understand the concepts of the semantic web, and I kind of know what RDF is, but my knowledge tails off shortly after this (I know that RDF triples exist, but not why they help) – what I need is RDF for Dummies or something.