My mother is a chronic Radio 4 listener and heard a little bit of a
programme the other day when in the car that she thought a friend
would be interested in. All she remembered was it was "about language
and culture". She had the mistaken idea that it was on in the morning.
She told me that she had been on the Radio 4 website looking to find
it but had no luck. Admittedly, it was quite broad search criteria.

I had a go at doing something about it today.
First thing I did was make a directory on my Mac, then ran the
following command:

curl -O http://www.bbc.co.uk/radio4/programmes/schedules/fm/2009/02/[01-28].xml

Then I tried grepping that data, but it wasn't pretty-printed and so
gave me the whole day's worth of programming for each result. So I ran
the following:

xmllint --format *.xml > combined.txt

This pretty-printed all the XML and wrote it out to a text file.

I then opened the resulting file up in MacVim. Here I had a
metadata-rich 33250-line text file containing details of all the
programmes broadcast on Radio 4 in the last month. I tapped "/" to
start a search and typed in "language". It took me to the
short_synopsis element of a programme element. I looked at the id, and
appended the relevant namespace on the front to give me
http://www.bbc.co.uk/programmes/b00hlcr2 - I loaded the page in my
browser, read the long description to my mother, who said something
along the lines of "Ooh, yes, that's right!". She then forwarded that
link and the link to the show page to her friend by e-mail. Total time
for me was about ten minutes. But the point was that she wasn't able
to do it herself - she had, as I said, gone through the listings pages
and the Radio 4 website and couldn't find it.

All this makes me very happy about the BBC's provision of excellent
metadata as XML, RDF, ASCII and HTML, without doing any silly API or
Web Services nonsense. It's great not only because people can build
applications on top of it, but just because nerdy people can find
stuff easier.

A suggestion for making this better: a sort of 'fuzzy' programme
finder - a very user-friendly search page linked to from iPlayer and
Listen Again (etc.) that would let you do natural-language searching
of programmes, sorted by recentness. So you could go on and select
that you saw something on TV or heard something on radio, maybe
specify a channel, maybe specify roughly when and throw it a few
keywords.

(I have to say, I did then just type 'language' into the BBC search,
and the first result in the TV & Radio Programmes box was the right
one. Having spent the last decade or so getting frustrated by the
*ahem* less-than-optimal search on bbc.co.uk, that's not the first
place I thought to look.)

-- 
Tom Morris
http://tommorris.org/
-
Sent via the backstage.bbc.co.uk discussion group.  To unsubscribe, please 
visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html.  
Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/

Reply via email to