- From: Joakim Söderberg <joakim.soderberg@ericsson.com>
- Date: Tue, 13 Jan 2009 16:43:04 +0100
- To: "Felix Sasaki" <fsasaki@w3.org>
- Cc: <public-media-annotation@w3.org>
I don't mean that we should use JSON but be inspired by what they do. But it is possible that your approach is a way forward. Maybe we can discuss it next time you attend the telephone meeting? /Joakim -----Original Message----- From: Felix Sasaki [mailto:fsasaki@w3.org] Sent: den 13 januari 2009 16:32 To: Joakim Söderberg Cc: public-media-annotation@w3.org Subject: Re: data interchange format Hi Joakim, I tried a different approach which is again not specific to a format (JSON / RDF / XML / ...). It is an update of the XSLT implementation, but that is replacable. I updated http://www.w3.org/People/fsasaki/annotation-mappings/ to contain a property "dateGeneral", in addition to "pubdate". "dateGeneral" encompasses both "pubdate" and other kinds of dates, e.g. "modification date". So querying a feed using the property "dateGeneral" gives you a superset of the results of "pubdate". See e.g. http://www.w3.org/People/fsasaki/annotation-mappings/query/q?inputURI=http%3A%2F%2Fgdata.youtube.com%2Ffeeds%2Fbase%2Fvideos%2F-%2FExperte%3Fclient%3Dytapi-youtube-browse%26v%3D2&property=pubdate vs. http://www.w3.org/People/fsasaki/annotation-mappings/query/q?inputURI=http%3A%2F%2Fgdata.youtube.com%2Ffeeds%2Fbase%2Fvideos%2F-%2FExperte%3Fclient%3Dytapi-youtube-browse%26v%3D2&property=dateGeneral I like the levels of granularity you describe DC > EXIF > XMP, but I'm not sure if you need a specific format to achieve that. The granularity description itself is enough to achieve the effect you want, see above implementation. And a format would create the burden that it must be implemented by tools who normally don't process Jason, e.g. browsers. Felix Joakim Söderberg さんは書きました: > Hello, > The way I see it, is that the definition of the data interchange format [1] is part of the API and therefore important. > > If we define a flexible format (like JSON) we could define type-value pairs or an array thereof which defines what you get (preferably in a simple way). It could solve the granularity problem i.e. "dc:rights vs. xmpDM:copyright" by informing what attribute is referred e.g. [Disney,dc:rights] [Walt Disney Company ,xmpDM:copyright]. > > We could define what a valid array should look like: > [(value, attribute), (value, attribute),..., (value, attribute)] > > - and valid values for "value" and "attribute" in BNF for example. > > The ontology could then perhaps define the levels of granularity e.g. (from top to bottom) DC -> EXIF -> XMP being the order of the elements in the array, similar to the schema of preference defined by the Metadata Working Group. > > Just some thoughts to get the discussion going... > /Joakim > > > [1] http://www.w3.org/2008/WebVideo/Annotations/wiki/Dataformat > > >
Received on Tuesday, 13 January 2009 15:43:47 UTC