backstage.bbc.co.uk

Use Our Stuff To Build Your Stuff

Prototypes

Missing Words Round

It's a prototype for an online version of the 'Missing Word' Round from 'Have I Got News For You'. It takes 100 most recent headlines from the BBC Backstage RSS news feed, updated every 15 minutes. The missing words are selected semi-randomly by the server, so may not have quite the comic potential of the TV version. PHP isn't renowned for its sense of humour.

The feeds are filtered for certain keywords, since death, murder and destruction aren't really that funny. The odd one may still slip through...

Because it is based on an RSS feed, there's the potential for 'guest publications' - just like the show. That is, assuming 'Knitting Machine World', 'Sea Shell Quarterly' or 'Centre Parting Monthly' publish an RSS version of their headlines!

To-Do


  • Add 'funniness voting' to previous user guesses. Can then give the 'funniest' answers on the homepage, or show them more prominently on the answers page

  • Find a better way of analysing the headlines so 'better/funnier' words can be blanked out

  • Include more RSS feeds, perhaps giving users a choice of which one

  • Give credit if you are close to the answer, rather than having to get it exactly

  • 01 Jun 2005 03:18 PM

Comments  Post a comment

I've seen some really innovative stuff on BBC Backstage, but this is the best so far!

I'm not very good at it, unfortunately, even though I subscribe to many of the BBC RSS feeds.

Any plans to open the source up for this so others can contribute to the engine? While playing it a whole bunch of suggestions came into my head, and I can't be the only one!

I'd certainly be happy to contribute some PHP if you were willing to release the sources.

  • 2.
  • On 02 Jun 2005 03:18 PM,
  • Kirk said:

Fantastic idea - I haven't had an odd one yet, although I did get the same one twice in a row.

Very good, and very addictive waste of time during lunch hour.

One issue that you might want to solve is that it occasionally shows the same headline again? Then again, if you're only sampling 100 perhaps not.

After using it for a little longer, I've noticed some odd things seem to be creeping in. These are generally headers for sections of the site - your pictures for example. I just got this one: 'Your Pictures of Scotland: 27 May - 3 June'

  • 5.
  • On 09 Jun 2005 10:51 AM,
  • nick said:

Thanks for the comments.

Kirk: Since its initial launch I should have solved the problem of getting the same headline more than once. It uses cookies though, so if they are blocked you might get them twice.

Angus: yes, the feeds do seem to include this type of headline too - I've filtered a lot of them out, but since the only way of knowing is when they appear, I just have to keep adding to my list of 'avoid' words!

Rich: I've nothing against opening the source code - it was built in a very quick and dirty way (in one day) so probably isn't an example of best practice! While most of it is fairly standard stuff (magpie to get the RSS) I thought it might be more useful to move some of the code into libraries and open source those -- e.g. the code for picking which words to blank out. At the moment its very crude, and I'm sure someone with more experience of natural language processing or AI could do a much better job.


Post a comment




Remember Me?




style: lo-fi | hi-fi