Note: this service is currently Alpha. Please send any feedback to channelography@rattlecentral.com.

About

Channelography was originally born out of a frustration with video and in particular with being able to index moments and things in video (as opposed to the entire piece of film). Then came a conversation, a brief one, with Steve Bowbrick and Jem Stone from the BBC when we were showing another piece of work. We got talking about Muddy and how you might be able to index closed captions and at that time they'd just been made available in iPlayer. Then somebody mentioned that it'd be nice to see what was really put out on the BBC and how that could form another representation for channels, brands, even programmes. That was a while ago.

Then we mentioned it to BBC R&D and they said "could you build that then". And we said "OK". And then a bit later we did build it. It was a bit of an experiment as we didn't really know how many programmes were on iPlayer, how many of those programmes had captions or indeed, what the issues might be with extracting terms from verbatim subtitle 'data'. I say 'data' but really this is people paid to listen to a programme and then type what is being said very quickly, often in real time and often with spelling mistakes and stuff. Childrens programmes particularly throw up some interesting problems because their use of language tends to be more playful. "Ing" is mentioned in a sentence that goes "Ing a Ling" and we referenced that as being the organisation ING Group (we don't now, the system learns from it's mistakes). As well as closed captions, we also used the wealth of data available on /programmes (such as genre and format 'type') to be able to say things about channels and going forward we hope to be able to bring in more data.

Anyway, we now know lots more about closed captions and what we can start to represent well and what we can't and how we go about making it better. At this stage we see it as being a stimulant for thinking about ways into and out of programmes and channels, as much as a way to represent channels themselves. We'll be tinkering with channelography further so please do let us know what you think and what, if any, ideas this has sparked. Thank you.