[vdr] Collecting semantic channel meta data in a way to be useful for VDR

Henning Pingel henning at henningpingel.de
Mon Jun 18 14:52:37 CEST 2012


 Hi,

 I posted in the discussion "RFE: Make VDR more friendly when using 
 combinations of DVB-S, DVB-T and DVB-C" yesterday and mentioned that my 
 little project Channelpedia could be helpful in this regard. But because 
 my approach is more targeting the necessary semantic meta data that 
 would be necessary for those advanced features, I would like to explain 
 my ideas in a different thread.

 When I started to work on Channelpedia my goal was to offer a website 
 where different VDR channel lists of different DVB providers could be 
 easily stored, sorted and regularly updated in an easy way. The main 
 audience were new VDR users who often are overwhelmed by the amount of 
 channels offered by their DVB source (in particular satellite sources).

 Technically, Channelpedia is a PHP web application that uses a sqlite 
 database and offers to upload channel data and retreive channel data in 
 different ways (static generated HTML pages + a small RESTful JSON 
 interface). Channelpedia is not a VDR addon and not a VDR plugin. The 
 coding is completely independet from VDR and it can't be plugged into 
 VDR in any way.

 Now that there are a couple of different DVB providers in the 
 Channelpedia database, I thought it would make sense to additionally 
 collect semantic meta data for channels in a provider independent way. 
 This semantic data can be gathered and stored in Channelpedia, but in 
 the end it should be easily exportable from Channelpedia. For example in 
 a single textfile, so that this textfile could be used by VDR plugins 
 (or even VDR core) for lookups.

 The basic idea is that this textfile has a certain structure (could 
 also be XML or JSON) and that anybody could create this data file by 
 typing it all down manually. But Channelpedia on the other hand could 
 create this file in an automated way so that it is rarely necessary to 
 maintain that file manually. At this time, Channelpedia collects some of 
 this meta data only for channels relevant for German language countries 
 (due to the amount of work involved I have limited it to German channels 
 for now).

 I will now give an example of the kind of metadata I would like to 
 collect per channel:

 All this metadata below is not specific to one single DVB provider. 
 Instead it should be usable for any provider (C,T,S,I,...) that offers 
 the respective channel.

 * Country/countries where the organization is based that offers
   the channel (problems might arise with cooperations: 3sat is a joint
   effort of Germany, Austria, Switzerland)

 * Primary region of the audience: Where does the indended audience 
 live?
   Which language do they speak?

 * Website URI of the channel (Wikipedia page of this channel +
   the website of the organization)

 * Type of channel (Kids entertainment, Music, Crime, Sport, ....)

 * Channel logo resources for this channel that can be found on the web

 * List of existing EPG web services (that offer an XML based
   epg for this channel)

 * Existing channel variants with the completely the same
   or nearly the same epg:
   - Time-shift-versions (+1, +2, +24): List of other channels that
     have the same epg but a timeshift offset

   - Regional variants (BBC 1 London, BBC 1 Oxford, ... /
     WDR Aachen, WDR Köln, WDR ....): List of other channels that
     share 95% of the EPG but have "regional windows"

   - Quality/Bandwidth variants (SD, HD, different bitrates,
     DVB-T offers less quality in Germany compared to SD via DVB-C/S):
     List of channel alternatives that offer a higher or lower picture 
 quality)

   - Reference to root channel (the "root" channel is
     the predecessor of all other variants)

 * CA-status: This channel regularly changes from FTA to
   scrambled or is permanently scrambled.

 * Which pay-TV packages / platforms have got this channel in their 
 portfolio

 To store all this metadata for a channel, the channel needs a unique ID 
 that is different from the technical unique ID that VDR uses internally 
 (Source-NID-TID-SID, for a more detailed explanation of this ID please 
 read VDR man page). The technical IDs of VDR are only used to identify a 
 channel within the channels.conf of one VDR user and his active DVB 
 providers. But we need IDs that are able to identify channels across 
 different providers. Therefore we have to make them up ourselves, based 
 on their channel name. I want to call them Channelpedia IDs for now - 
 short: CPID, to make it easier to talk about them.

 But the NID-TID and SID still remain very important. If we put all the 
 NID-TID-SID data into the metadata too, we can use them to automatically 
 assign a channel in a channels.conf to a CPID.

 Regarding German channels, the DVB-C sometimes reuse the same 
 NID-TID-SID of the channel on satellite. This means that we can hope 
 that we don't have to store as many variants of NID-TID-SID in the 
 metadata as there are providers out there.

 I guess this metadata collection can never be complete or completely 
 accurate, but even if it doesn't contain all those channels out there it 
 could still be very helpful for VDR plugins. Example:

 I create a searchtimer using epgsearch and epgsearch is able to look up 
 all channel siblings in my local channels.conf that share the same EPG 
 with the channel that I have currently selected. Now, epgsearch could 
 give me choices: "Do you want to record the HD version of this channel? 
 Is SD quality good enought? etc."

 Further infos:
 <http://channelpedia.yavdr.com/>

 Demos:
 <http://channelpedia.yavdr.com/gen/de_uniqueIDs2.html>
 <http://channelpedia.yavdr.com/gen/de_uniqueIDs.html> (JSON files 
 available for auto-assignment)

 German language thread: 
 <http://www.vdr-portal.de/board1-news/board2-vdr-news/111607-announce-entwurf-channelpedia-generiert-automatisch-unique-kanal-ids/>

 Regards,
 Henning



More information about the vdr mailing list