Need a new entity to represent a "media depiction of an event/action/etc"
bquinn opened this issue · 8 comments
What we have described so far is the actual on-field actions, team structures etc.
Thinking about a media context, we want to be able to link a news item, video stream, broadcast feed etc to an Action, or an Event.
Obviously we may have multiple broadcast depictions of the same action.
We need to represent the date/time/timecode of the media separately from the action itself - even the "live feed" of a sport (eg a golf event) will broadcast footage of a shot delayed by a few minutes much of the time, because the action happens simultaneously. See #148 .
Questions:
- What to call this new entity?
- What properties does it need? broadcastDateTime for video, what does it need for non-linear news like a text news article?
- What kinds of things could it link to?
Some notes from today's meeting:
SNaP Ontology has the concept of "tags" which is somewhat related
schema.org has https://schema.org/recordedAt / https://schema.org/recordedIn which links Event and CreativeWork. So this is similar to what we want to do here.
Some early thoughts:
We could look at EBU core as well:
https://tech.ebu.ch/metadata/ebucore
So something like...
Example: video of a goal in football
<Action123> rdf:type sport:Action . # Ronaldo kicking a goal
<Annotation123> rdf:type sport:Annotation .
<Annotation123> sport:subject <Action123> . # target is Action123
<Annotation123> sport:annotates <https://championsleague.com/video123.mp4#t=4560,4570> # uses media fragment to point to the exact point in the video where the action begins and ends (unit is seconds)
Thoughts from today's call:
- Do we want to link straight to the video with the media fragment included in the URI, or should we create an intermediate entity/object to describe the video and then have a property to define the start and stop time(s)? Our initial thought is that the latter would be too cumbersome to model and process.
- What about video with different renditions / versions? eg 1080i vs 720p renditions of the same video track. Is there an idea of a generic "parent video" that can have the media fragment attached, and then "fall back" to the appropriate version when requested by a client? (Maybe this is a good question for the SVG folks, or DPP)
- Do we define our own entity for Annotation (and therefore "subject" and "annotates" properties) or use the W3C annotation ontology definition? - this is the main question we need to answer right now before we move forward
- How to handle video about things that we don't necessarily describe in SportSchema, like the celebrations at the end or the coin toss at the start? Maybe we just say they are out of scope?
Example: news story on a web site about an event (eg a football game)
<Event123> rdf:type sport:Event . # The overall football game in which Ronaldo kicked a goal
<Annotation123> rdf:type sport:Annotation .
<Annotation123> sport:subject <Event123> . # target is Event123
<Annotation123> sport:annotates <http://newsoutlet.com/article-about-football-game.html>
Example: news story on a web site about a competition (eg Champions League 2023 season)
<Competition123> rdf:type sport:Competition . # Champions League 2023 season
<Annotation123> rdf:type sport:Annotation .
<Annotation123> sport:subject <Competition123> . # target is Competition123
<Annotation123> sport:annotates <http://newsoutlet.com/article-about-champions-league-2023.html>
What about video with different renditions / versions? eg 1080i vs 720p renditions of the same video track. Is there an idea of a generic "parent video" that can have the media fragment attached, and then "fall back" to the appropriate version when requested by a client? (Maybe this is a good question for the SVG folks, or DPP)
Suggest we leave that to the implementor? Many different patterns and a number of standards we could leave it open?
Do we define our own entity for Annotation (and therefore "subject" and "annotates" properties) or use the W3C annotation ontology definition? - this is the main question we need to answer right now before we move forward
Suggest IPTC define this. It seems like a key join across a lot of IPTC standards that it would be useful. As well as what is out there currently is not necessarily that focused.
Examples look good.
Still undecided - need to look into the W3C standards and see if there's something we can simply use, or whether we should import/define properties in sport schema.
Today we concluded that as Silver is recommending that we do this in a generic way across IPTC standards, we probably want to do it outside of Sport Schema.
Therefore, we don't have to do it for the Sport Schema 1.0 release.
So we have removed this issue from the 1.0 milestone.