isn't it more about modelling writers, storylines and fictional
characters. Imagine a pvr that you could set to only record programmes
featuring the character Sarah Jane Smith (whether dr who or SJA)?
Probably quite a way off tho

Anybody using the Digiguide program - www.digiguide.com - on a PC to
drive a TV capture card (with the help of this plugin -
http://www.spidersweb.freeserve.co.uk/dab/dabdig.htm) is not that far
away from that. I've been using for about three years to capture
programmes on the basis of who was involved (ie, I don't normally watch
QI, but I have a marker set up for Rich Hall the comedian, and I get any
episodes of QI, or any other programme, that he is involved in. Though
obviously this is reliant on having the right metadata - at the minute
it's on the basis of the text description, but it could be much better.
 
There's no reason why Freeview recorders or Sky+ couldn't do this,
beyond getting the user experience right. 
 
Aaron
 


________________________________

From: [email protected]
[mailto:[email protected]] On Behalf Of Michael Smethurst
Sent: 08 January 2009 12:48
To: [email protected]
Subject: Re: [backstage] Your ideas are now finally welcomed





On 8/1/09 12:09, "Frankie Roberto" <[email protected]> wrote:



        
        On Wed, Jan 7, 2009 at 4:00 PM, Michael Smethurst wrote:
        
        

                you're not wrong. we do do a fair bit of user testing
around this stuff and attempt to match what we model to audience mental
models. it's not always easy tho. we also need to consider how
programmes are grouped outside of web pages - eg series record on pvr,
podcasts etc. for all your examples below consider what happens when u
press series record - what do you expect to be recorded?
                

        
        Sounds good. I hadn't thought about the series record button.
(It's only recently been added to Freeview, or at least my PVR has only
recently been updated to use it - it's so so useful, I used to have to
set up recording times based upon search strings, so thanks to whoever
got that to work.)
        
        it is very sweet. It's also handy to have to get your mind away
from web pages and get a different perspective for modelling
        
        

                > Part of the problem is that it can all be so
complicated, and there are tons of edge cases!
                
                more edge cases than not. knowing when to stop modelling
is the tricky part
                

        
        Indeed, and if the [relative] success of microformats vs
Semantic Web has taught us anything, it's that simple can be better.
        
        cough
        
        

                > What's the relationship between Doctor Who and Doctor
Who Confidential? They are different programmes, but very closely
related (part of the same brand, but also a 1:1 relationship between
individual episodes).
                
                we'd say they're 2 different brands grouped by a
franchise. series record dr who and u don't get DWC! so far we haven't
modelled episode > episode relationships (episode <depictsMakingOf>
episode) but franchises (horrible word but the best we could come up
with) are coming soon
                

        
        Sounds good from a modelling perspective, pity the language is a
bit counter-intuitive. For me, 'Doctor Who' is the brand, DW/DWC/etc are
the 'shows', and the individual programmes are 'episodes'. But I guess
it doesn't matter too much. Interesting question is whether
Torchwood/Sarah Jane Adventures fits within the Doctor Who 'franchise',
or is linked in some other way (Freebase has the concept of 'fictional
universes' which I quite like).
        
        isn't it more about modelling writers, storylines and fictional
characters. Imagine a pvr that you could set to only record programmes
featuring the character Sarah Jane Smith (whether dr who or SJA)?
Probably quite a way off tho
        
        
         
        

                > How about the Doctor Who Christmas specials? Which
series do they fit into, the beginning of a new series or the end of the
last one? Or are they their own 1-part series?
                
                they belong directly to the doctor who brand. ie they're
not in any series. pips supports a mixed content model so a brand can
have a combination of child series and child episodes
                

        
        Nice. Although it'd be good if you could capture the chronology
in some way (the fact that the such-and-such Christmas special fell
between series 2 and 3). You could rely on the first broadcast date as a
workaround though, I guess.
        
        ordinality is stored where known (ie for recent programmes - see
our previous conversations on get satisfaction about episode guides
around exactly this). So we can say this episode sat between series 2
and series 3
        
        

                > What about the Doctor Who Children in Need Special
(apologies for all the Doctor Who examples) - should they fit under both
the Doctor Who and Children in Need brands, or just one or the other?
                
                NEVER mention CIN around BBC data modellers. it makes
them whimper
                

        
        Heh. The show-within-a-show problem that Children in Need
presents crops up more often than you might imagine. 
        
        Trust me - every one of them is on a whiteboard somewhere around
here
        
        I think kids tv shows, Comic Relief, and maybe T4 (?) and the
old Big Breakfast show all do it. It always frustrates me when the TV
listings have one big long show down in the schedule, but I only really
wanted to watch a show within it.
        
        programmes within programmes are everywhere.  Even thought for
the day kinda fits the definition. The trouble with CIN and CR is the
episodes are specially commissioned as part of the campaign. The CIN dr
who doesn't fit into the narrative arc of dr who proper. It's both CIN
and DW but it's MORE CIN than DW. Press series record on DW within CIN
and what gets recorded?
        
        
         
         
        

                All of these changes will also be make to the programmes
ontology very soon
                http://purl.org/ontology/po/
                

        
        Look forward to it. Any chance of making the document a bit
easier to understand whilst you're at it? (I had to read it half a dozen
times before I understood, and even now I'm not 100%). Some concrete
examples of shows would be a help for starters. I think half the problem
with the Semantic Web is that it's so damn difficult for people to
understand (even web developers...).
        
        The new version should be a little cleaner. Possibly best to
look at the rdf on /programmes for examples
        
        Have you thought much about how it applies to radio btw? The
structure of radio shows seems to be quite different from TV. 
        
        We work in radio and music so yes. It's not SO different
        
        One pet bugbear of mine is the way that 'stand-in' presenters
are dealt with. It might be presenter A's usual slot in the schedule,
but if presenter B is filling in, then it's really presenter B's show.
Yet within listings and the now/next data it's treated as if it was
still presenter A's show. You even get the madness of presenter B
telling listeners to e-mail in by sending to presenter A's e-mail
address! (This problem is the reason why my Radio Pop profile shows me
as having listened to quite a few of George Lamb's shows - in reality I
was listening to his [much better] fill-ins).
        
        Stand in presenters are a nightmare. Trying to get radio
production people to disentangle the programme from the presenter when
they share the same name is difficult. You sometimes end up with the
guest presenter's photo sat alongside the usual presenter's name cos
that's the name of the programme. We are trying to figure out ways of
fixing this and properly modelling presenters etc. 
        
        

                
                

        
        
        
        



Reply via email to