Bibliography

True to my word, my bibliographic database as of today is linked from the sidebar. Lots of Semantic Web, inference, Web, etc. stuff, as well as some general Computer Science. Links are provided where I've put them into my database, and the raw BibTeX is hiding behind the scenes too.

Posted at 2004-04-30 16:41:56 by Richard • Link to Bibliography
Comments, trackbacks.

BibTeX

I was talking to Mike today about bibliographies, and why nobody bothers to put theirs on the Web.

The answer: tool support. I've got a database of 200-odd papers, articles, and websites. Ones that need a Web reference have a \url{} entry in the Note field. The tool I use (Bibdesk) is also crap at standards, so I have to manually strip the curly brackets from around numbers, do the LaTeX special characters, and so on.

Then I come to generating stuff for the Web.

My ideal would be a pretty, XHTML 1.1 page with URLs and local file links. Even better if it's got some kind of smarts (pages per author, customisable style sheet, etc.).

Instead, we have:

hevea

Converts TeX to HTML. Does OK, but spits out HTML 4.0 Transitional with big ol' FONT tags, no hyperlinks, no file links, etc. (partially because it takes the data from the presentation file, not the bibliographic database).

bibtex2html

Big ol' package, which sort of does the job. It prints out my notes, so links come out twice — my problem to fix. However, it spazzes up the annotation/abstracts (linebreaks? what linebreaks?).

What I'd like is a really intelligent parser that will use BibTeX's implicit ontology to generate OWL-DL which I can then query and format using some standard tools. Then I'll wake up.

Ah hell, I'll just put up the plain pages.

Posted at 2004-04-30 15:30:33 by Richard • Link to BibTeX
Comments, trackbacks.

Fair measures?

Here's something to ponder before you put away your tin foil hats. Last time I filled up with petrol, I did so at Tesco (Clubcard points!). 30mpg — just over, in fact.

Now, although I did do marginally more town driving with the last tank, I just got a little over 26mpg, and I managed to fit a few more litres in the tank than I was expecting. It just occurred to me: how do we know that when the petrol pump says 52 litres that it's actually dispensed 52 litres? Perhaps I only put in 48 litres, and the pump was lying?

Are there industry bodies to check them, as with market trader scales? Are some more accurate than others? Is the fact that I filled up at Total a factor in the perceived mileage my car gets, and the very real amount that I pay for a fill-up? Should I have gone to Tesco?*

* Yes — Total was 80.9p/l, and the bonus card isn't as good.

Anyone else think that paying £40 for a tank of petrol is reasonable? I'm always startled when Americans complain about their petrol prices, and we have to explain that the price I just said is per litre, not per gallon, and that it's pence, not cents!

Posted at 2004-04-30 14:17:22 by Richard • Link to Fair measures?
Comments, trackbacks.

Prima Facie

Bijan Parsia makes an argument in favour of the Semantic Web: namely, using URIs as a specific sort of hypertextual link between the subject and the object*.

Of course, this rings a bell for me (no, silly: not the 200 Semantic Web articles and papers I've read in the past 6 months!) — Tim Berners-Lee's original view of the Web (or the Mesh as he then termed it) was as a network of typed nodes (e.g. a project, or a code module, or a person, or a document) with typed links (e.g. worked on, supervises, etc.).

I.e. the Semantic Web.

Not bad for 1989, eh?

* of course, plain RDF is no good for this: we need RDF Schema (or preferably OWL) to actually accord some meaning to the links through shared ontologies. Remember: RDF is just a way of describing triples. RDFS brings classes and property inheritance; OWL adds more advanced ontological semantics, like class disjointedness.

Posted at 2004-04-30 13:16:49 by Richard • Link to Prima Facie
Comments, trackbacks.