Ivo Emanuel Gonçalves
2007-Sep-14 21:59 UTC
[ogg-dev] The use for an XML based metadata format
I've been trying to follow the situation here, because that's one of the things I try to do in Xiph: know what's happening. There's a total of 70 messages related with recent metadata discussions. Here's what I have understood so far: * Daniel is still not trying to read and understand the _other_ metadata proposals; he makes several non-true statements regarding them. * Daniel wants to "stuff in" a new metadata format no matter the cost to backwards-compatibility; he also intends it to completely obliterate Vorbis Comment. All of this is highly ambitious, especially for someone with no actual knowledge in the field. * There's interest again in metadata in Ogg. This be a good thing. * Skeleton still seems to me to be the more advanced and logical step for metadata in Ogg. Here's my take on the issue: Metadata is highly overrated outside the Semantic Web. As far as I see, that's why nobody actually cared to support anything but Vorbis Comments, ID3 hacks (in MP3), and APE tags (in whatever proprietary format that uses that). I believe Matroska has no metadata support either, but I may be wrong. So, if it's for the Semantic Web, using W3C-based proposals are likely the best solution. After all, they are the ones behind the whole thing to begin with. RDF is probably too complicated for Ogg, though. I guess in the end I can't offer any decent help or advice right now, but I felt like writing this anyway. How does that American saying goes? "It felt like a good idea at the time". -Ivo
Ivo, Let's not start finger-pointing, since that's not constructive. I do think we have gotten further than what you are describing. Skeleton, CMML and Vorbiscomment are not sufficient for some people - in particular when there is a more complex piece of metadata to be communicted, e.g. from an archive that stores a lot more information than just name-value pairs. But I agree: the space is complex and creating a good, generic description scheme for e.g. music alone is hard enough, not to speak of the complexity of a generic, but structured annotation scheme for audio and video. I think we are focusing now on particular aspects of the problem and that is a lot more constructive. E.g. adding an ID to skeleton is a good thing. Creating a automatically parsable scheme that describes relationships between logical bitstreams is also a good thing. If you have constructive contributions, go ahead - else, just watch. Cheers, Silvia. On 9/15/07, Ivo Emanuel Gon?alves <justivo@gmail.com> wrote:> I've been trying to follow the situation here, because that's one of > the things I try to do in Xiph: know what's happening. There's a > total of 70 messages related with recent metadata discussions. > > Here's what I have understood so far: > > * Daniel is still not trying to read and understand the _other_ > metadata proposals; he makes several non-true statements regarding > them. > * Daniel wants to "stuff in" a new metadata format no matter the cost > to backwards-compatibility; he also intends it to completely > obliterate Vorbis Comment. All of this is highly ambitious, > especially for someone with no actual knowledge in the field. > * There's interest again in metadata in Ogg. This be a good thing. > * Skeleton still seems to me to be the more advanced and logical step > for metadata in Ogg. > > Here's my take on the issue: > Metadata is highly overrated outside the Semantic Web. As far as I > see, that's why nobody actually cared to support anything but Vorbis > Comments, ID3 hacks (in MP3), and APE tags (in whatever proprietary > format that uses that). I believe Matroska has no metadata support > either, but I may be wrong. So, if it's for the Semantic Web, using > W3C-based proposals are likely the best solution. After all, they are > the ones behind the whole thing to begin with. RDF is probably too > complicated for Ogg, though. > > I guess in the end I can't offer any decent help or advice right now, > but I felt like writing this anyway. How does that American saying > goes? "It felt like a good idea at the time". > > -Ivo > _______________________________________________ > ogg-dev mailing list > ogg-dev@xiph.org > http://lists.xiph.org/mailman/listinfo/ogg-dev >
On 15 Sep 2007 at 16:53, Silvia Pfeiffer wrote:> Skeleton, CMML and Vorbiscomment are not sufficient for some people - > in particular when there is a more complex piece of metadata to be > communicted, e.g. from an archive that stores a lot more information > than just name-value pairs.In my opinion, for this purpose it would be more appropriate to store the data separate from the audio files -- perhaps another file in the directory, or in a database. -- -:-:- David K. Gasaway -:-:- Email: dave@gasaway.org -:-:- Web : dave.gasaway.org
On 15/09/2007, Ivo Emanuel Gon?alves <justivo@gmail.com> wrote:> Here's my take on the issue: > Metadata is highly overrated outside the Semantic Web. As far as I > see, that's why nobody actually cared to support anything but Vorbis > Comments, ID3 hacks (in MP3), and APE tags (in whatever proprietary > format that uses that). I believe Matroska has no metadata support > either, but I may be wrong. So, if it's for the Semantic Web, using > W3C-based proposals are likely the best solution. After all, they are > the ones behind the whole thing to begin with. RDF is probably too > complicated for Ogg, though. >I suspect one reason this hasn't been done before is that it's much harder to do this kind of rich metadata (for reasons of specifying the structure, generating the data and incorporating it in the bitstream) than to do the quick and dirty 'tags' that most common media files use in their various incarnations. As Ralph Giles pointed out a while ago there are levels in metadata and we see all of them in operation: from the free-form social tagging to highly structured rdf. Whatever we do should be easily mappable to existing schemes at the least and this is also why I'm behind making room for using existing formats; what we want, certainly in the web context, is to plug easily into existing schemes, otherwise it's a no-starter for that sector. More to follow in a minute, but it's different enough to need another email. -- imalone