[mythtv-users] Inconsistent program naming in EIT guide on FreeView (UK DVB)

[junk]

Nick Morrott wrote:
> On 27/02/07, junk <junk at giantblob.com> wrote:
>   
>> Thanks. I wonder if you're having more luck because of the guide you're
>> using or the type of scheduling. Does the 'find and record one showing
>> of this title each week' setting, simple ignore the extended program
>> description and assume all showings in a given week are the same episode
>> or is it more intelligent than that?
>>     
>
> (I think) you need to acquaint yourself with how MythTV matches
> duplicates. When scheduling a recording, you can opt for no dupe
> matching, use the subtitle, use the description, use both subtitle and
> description, or (in latest SVN) use subtitle then description. The
> dupe checking alone should ensure you do not get (excessive) multiple
> copies of the same episode of a series. How well this works obviously
> depends on how accurately data for subsequent showings matches. In my
> experience of the RT grabber, it works pretty well.
>
> What dupe checking are you currently using for Ugly Betty?
>   
Subtitle and description. I can't say for certain whether the EIT 
information is inconsistent across the duplicate showings as I don't 
have access to the machine right now but when I checked a load of 
duplicate Two Pints... episodes last week, the subtitle/description 
fields were often different (sometimes different punctuation, sometimes 
typos etc.). So it's not Myth's fault and I don't think it's the 
settings I'm using, it's the EIT guide not being consistent.
> Checking my current RT data for Ugly Betty, all of the
> titles/subtitles (except for episode shown during T4 on Sunday) and
> descriptions (except for reviewed episodes) match across repeated
> showings. This enables the scheduler to intelligently identify which
> episodes are 'the same' even across channels and should results on
> only one recording of an episode.
>   
It sounds like Radio Times has more consistent subtitle/description for 
these showings. FrewView EIT is full of typos and situations where 
different people have obviously filled in the details for different 
showings of the same episode and have had different ideas about what 
should go in which fields of the guide (e.g. episode title or number 
sometimes in subtitle, sometimes in description, description sometimes 
has quotes around it, etc. etc.)
>> FWIW, I noticed, browsing through the Myth sources, that the EIT guide
>> scanner scanner goes to some lengths to match duplicate programs in the
>> face of inconsistent guide data - it seems to score programs based on a
>> count of words in common between the two showings - but the recording
>> scheduler does a dumb textural compare. (specifcially eit.cpp,
>> score_match() versus ProgramInfo::IsSameProgram()). Can any of the
>> developers comment on why this is done differently and whether the
>> scheduler would benefit from using the same heuristic?
>>     
>
> Not a developer, but for interest there is also a new technique for
> EIT data whereby programid/seriesid which are broadcast in the EIT
> data can be used for program identification and duplicate detection
> (see http://cvs.mythtv.org/trac/ticket/2811).
>   
That might be worth trying. However, if the people inputting this data 
are as sloppy as the people that fill in the text parts of the guide 
then it might not help! I should maybe just switch over to using Radio 
Times via xmltv if the guide data is more consistent.

Thanks
-- jeek
> Nick
>