To assert that Peter Sefton created this resource, wrap this link around his name in the text: http://ontologize.me/meta/?r=http://purl.org/dc/terms/creator&o=http://trove.nla.gov.au/people/541658The same technique could be used to refer to scientific terms, or proteins, or data sets, but with an explicit statement of what the link means. I will go through some of my thoughts about the Beyond the PDF workshop, describe this approach in more detail and finish up with a demo.
- EPub which is a zip-based format with HTML inside, has the ability to load-in any kind of extended content you like, so is a potential candidate for re-flowable content plus supporting data, plus, even a PDF. EPub is supported on all major platforms including mobile devices.
- HTML 5 manifests also let you make packages – so an article page in a journal can include a list of all the stuff your browser needs to download if you decide to ‘Save as App… ‘ – again, including all sorts of data and visualisations, even annotation clients and, of course, PDF. I linked to a simple demo that my team at ADFI put together in my last post.
- Capturing and identifying data – so it can be referenced in papers, theses, blog posts, emails, etc.
- Capturing and identifying the documents we’re working on.
- Labelling and describing the above using unambiguous metadata frameworks.
- Packaging all of the above using some abstract model that captures relationships between different resources.
- Maybe, in some disciplines where there is a demonstrable ROI, being able to embed machine-readable semantics in publications.
- I’m an Australian researcher.
- I use a word processor to write my papers.
- I tend to blog works-in progress, and I’m working in a domain where my impact depends on this. That is, blog posts are an important form of scholarship.
- I do write scholarly articles, often they start life as conference papers and are then turned into journal articles later.
- Use my people Australia ID minted by the National Library of Australia. http://nla.gov.au/nla.party-541658 . Yes there are other forms of ID but it’s widely recognised that the major providers are going to work in a distributed way so if I use this one – if I get an ORCID, the the NLA.
- Say that the person identified by “ http://nla.gov.au/nla.party-541658” is the author.
- Have my blog (and other downstream services) understand and advertise that I am the author.
By Peter SeftonNow, I can link that to People Australia.
By Peter SeftonDone! Actually, no. All I have done is add a link – good enough for an English speaker to work out that I’m the author, and to provide a nice identifying endpoint. So what I’d really like to be able to do here is make the relationship explicit using semantic web technology. I head off to the RDFa primer. Looks like I have to do something like this:
This is way better than my first attempt, and it seems to work at the Sindice RDFa inspector; people on the list have been helping me, particularly Paul Groth. But how do I do that in a word processor? I don’t. It’s just not supported. Even experienced HTML wranglers would have trouble with stuff if they had to do it by hand. Hey, what if I could go to People Australia and on that page it said something like:
<span xmlns:foaf="http://xmlns.com/foaf/0.1/" xmlns:dc="http://purl.org/dc/elements/1.1/" property="dc:creator" rel="foaf:maker" resource="
To assert that P M Sefton is the author of a resource, use this link: http://trove.nla.gov.au/people/541658?prop=dc:creatorThen in my word processor I could use that link. I could add something to my WordPress site so that when I send it links like http://trove.nla.gov.au/people/541658?prop=dc:creator it would generate the RDFa for me. And OJS could look for links like that, and the local ePrints site, and so on. But People Australia doesn’t support that (yet) – it’s not really the done thing to add extra attributes to a URL that resolves to someone else’s site. What if there was a site somewhere else, though, that let me generate URLS like that? As far as I know there is no such ‘real’ site, but I do have a demonstration of what it might look like. Try this:
http://ontologize.me/meta/?r=http://purl.org/dc/terms/creator&o=http://trove.nla.gov.au/people/541658?prop=dc:creatorThat’s my (clumsy) attempt to encode a triple in a link. My demo service is pretty basic – and doesn’t do content negotiation, although Duncan Dickinson did do a better version of the service.
|Subject||<The referring page>|
|Predicate / property||dc:creator|
|Format (not implemented)||http://purl.org/triplink/v1|
- Define an extra parameter in the nanoformat so that these kinds of links are identifiable, whether they resolve to Ontologize.me or some other site. I will probably register a PURL for this, the links would be identifiable by
&triplink=http://purl.org/triplink/v1or the like.
- Work out the syntax of the nanoformat in more detail with input from RDF experts.
- Most importantly, make tools.
- Word processor plugins that make this transparent. I have argued that the the MS Word ontology plugin could use this to make the tool (a) much more interoperable, and (b) add relations to the mix – at the moment it links text to ontological terms with no predicate, giving no more semantics than a plain old link.
- Websites for ontologies and name authorities that provide meaningful links people can use to express useful relationships.
- Plugins for content management systems, PDF readers etc. But note that even without plugins these ‘triplinks’ are usable, you can click them and have them resolve to something useful, and harvesters could look for them in HTML, Word, PDF, email etc and process them.
- Services like SWORD and repositories to look for link-encoded semantics when you upload stuff to them.
Sefton, P. et al., 2009. Embedding Metadata and Other Semantics in Word Processing Documents. International Journal of Digital Curation, 4(2). Available at: http://www.ijdc.net/index.php/ijdc/article/view/121 [Accessed October 22, 2009].
Copyright Peter Sefton, 2010. Licensed under Creative Commons Attribution-Share Alike 2.5 Australia. <http://creativecommons.org/licenses/by-sa/2.5/au/>