Wikidata:Property proposal/YouTube video

YouTube video edit

Originally proposed at Wikidata:Property proposal/Creative work

Not done

Description	YouTube video this video is published to
Represents	YouTube video (Q63412991)
Data type	Item
Domain	video and/or audio work (Q110879282)
Allowed values	YouTube video (Q63412991)
Example 1	Butter (Q106928663) → Q110911269
Example 2	Butter (Hotter Remix) (Q107025366) → Q110911271
Example 3	After We Ride (Q108214081) → Q110911274
Robot and gadget jobs	User:BorkedBot could create video and/or audio work (Q110879282) and YouTube video (Q63412991) items from current YouTube video ID (P1651) statements and link them together. Then, it could keep track of views over time using number of viewers/listeners (P5436) qualified with point in time (P585) on YouTube video (Q63412991) items.

Motivation edit

I want to keep track of video view counts over time and having dedicated items for YouTube videos and a property to relate them to their main video item that also links to other platforms can do this. The current usage of qualifiers on YouTube video (Q63412991) cannot keep track of view counts for more than one date.

@BrokenSegue: since they manage BorkedBot.

Discussion edit

Comment there's already YouTube video ID (P1651). Ainali (talk) 03:26, 8 February 2022 (UTC)[reply]
Oppose per Ainali. --Tinker Bell ★ ♥ 20:48, 9 February 2022 (UTC)[reply]
- @Ainali @Tinker Bell this property links video items and their dedicated YouTube video (Q63412991) items. Lectrician1 (talk) 02:55, 14 February 2022 (UTC)[reply]
  - Then, Q110911269main subject (P921)Butter (Q106928663). No need for a special property, any service can use the same way to express this relationship. --Tinker Bell ★ ♥ 03:30, 14 February 2022 (UTC)[reply]
    A YouTube video (Q63412991) is a software resource (Q110832782) and main subject (P921) should be used to describe the relationship between a work (Q386724) and an entity. We shouldn't use main subject (P921) to describe everything that has a descriptive/depicting relationship, especially since the relationship between software resource (Q110832782) and video and/or audio work (Q110879282) is a bit shady already. A dedicated property has the benefit of constraints as well and will help make sure that there are separate items for videos and YouTube videos. Lectrician1 (talk) 17:33, 14 February 2022 (UTC)[reply]
Comment Hmm, can we find better examples? The three example values does not seem notable and should probably be deleted. We can't create items for every YouTube video, no matter how much you love the artist. Ainali (talk) 09:25, 14 February 2022 (UTC)[reply]
@Ainali Why can't we? If the video is notable, we should be able to. We already store this data in qualifiers... Lectrician1 (talk) 11:58, 14 February 2022 (UTC)[reply]
Why? Because just being a YouTube video is not making something notable. I am not saying YouTube videos cannot be notable, but if they don't have any external references outside of YouTube (like all the examples), and no items are linking to them (like all the examples), what makes them notable? Ainali (talk) 16:14, 14 February 2022 (UTC)[reply]
@Ainali YouTube video view counts of music videos at a certain point in time/view-number achievements are considered notable and are often part of Wikipedia articles. You can see an example usage of this data for Q110911269 here. Therefore, we should be able to have them on Wikidata. Lectrician1 (talk) 16:27, 14 February 2022 (UTC)[reply]
Your claim about view counts rendering notability is not supported by Wikidata:Notability. Where did you find this "consideration"? Also, please remember that policies don't carry across projects (they often don't even carry across language versions of the same project). Ainali (talk) 20:48, 14 February 2022 (UTC)[reply]
@Ainali Well then why is the current usage of of placing such data as qualifiers for YouTube video ID (P1651) considered notable enough to be on Wikidata? I'm not introducing anything new, I'm just establishing a different way to structure already-existing data. Lectrician1 (talk) 21:04, 14 February 2022 (UTC)[reply]
The key is that your "such as" is a false analogy. By that measure, we could have an item about every tweet thanks to Twitter post ID (P5933). It only works if the external identifier is issued by some authority that has a curated selection and is not an identifier generated on a platform where anyone can publish. Ainali (talk) 21:47, 14 February 2022 (UTC)[reply]
@Ainali This logic still doesn't seem to hold up though. Anyone can still make statement about any Twitter post ID (P5933) or YouTube video ID (P1651) and put it on an item and add further information like User:BorkedBot does. If that is valid, why is it not valid if that information is moved to an item? I don't think any information that's not notable should be stored on Wikidata, regardless of how it's stored. Why is it being on an item suddenly not notable? Lectrician1 (talk) 00:31, 15 February 2022 (UTC)[reply]
I am afraid we are talking past each other. I am not saying that adding a property like this to a valid item is wrong (hence a comment and not a vote in opposition), I am saying that all the items used as values in the examples should be deleted and that this proposal could be improved by having properly notable items in the examples. Ainali (talk) 10:25, 15 February 2022 (UTC)[reply]
@Ainali And I'm arguing that your statement that the items are not notable and should be deleted is not valid. Please look at the items used as values. They stored the same information as they did when the data was stored as qualifiers on the main video item. If the data on the item is not notable, why is the original data stored as qualifiers notable? Lectrician1 (talk) 16:39, 16 February 2022 (UTC)[reply]
Are you arguing that any statement creates notability for an item? That would spiral out of control quickly and is not described at all in the notability guidelines. To me, it seems obvious that it cannot be true because then we should have items for every value and string, which is just ridiculous. Ainali (talk) 18:07, 16 February 2022 (UTC)[reply]
@Ainali So you're saying that it's perfectly fine to add data in the form of a qualifier right now, even if it's not notable... Lectrician1 (talk) 23:46, 16 February 2022 (UTC)[reply]
@Lectrician1 I am saying that we don't have notability on statement level. Which also means that just because something can be added as data in a statement, the same data may not be grounds for item notability. That does not imply that just because a property exists, all data can be added. We have constraints and abuse filters to protect from some data being added and some data may also fit better as tabular data on Wikimedia Commons. Ainali (talk) 09:06, 17 February 2022 (UTC)[reply]
@Ainali So these qualifiers are considered notable but these statements are not? Lectrician1 (talk) 14:10, 17 February 2022 (UTC)[reply]
No, you're getting it backwards. It's not the statements per se that is the problem in your second example, it's the mere existence Q110911269. If that item had been notable, the statements would have been fine. Ainali (talk) 22:24, 17 February 2022 (UTC)[reply]
Regardless of whether these YouTube videos should have their own item, it seems overly specific to have a property scoped for what is essentially one possible distribution platform of many. This is basically a manifestation of the work. There should perhaps just be a general way to create those linkages if there is not already, but it seems similar to has edition or translation (P747), if it can be extended to this meaning. Dominic (talk) 16:26, 15 February 2022 (UTC)[reply]
@Dominic I see the similarity too and agree that it is a manifestation, but it is a bit of a different relationship than has edition or translation (P747). YouTube videos are consistent data objects and books (which has edition or translation (P747) is mostly used for) are not consistent. I wouldn't be concerned if we do end up having dedicated properties for other video sites as they too each are unique in their data objects and there's honestly not that many out there (most videos are just posted to YouTube). Having a dedicated property also works nicely with the benefit of constraints. Lectrician1 (talk) 16:44, 16 February 2022 (UTC)[reply]
why not instead just use number of viewers/listeners (P5436) and add a qualifier for the youtube video in addition to the date? BrokenSegue (talk) 03:53, 16 February 2022 (UTC)[reply]
@BrokenSegue We already do that. However, I want to store multiple timestamps of view counts for YouTube videos and to do that we require an item. We cannot do it with qualifiers. Lectrician1 (talk) 16:33, 16 February 2022 (UTC)[reply]
I'm not following why my proposed solution won't work for multiple timestamps of view counts. BrokenSegue (talk) 16:49, 16 February 2022 (UTC)[reply]
@BrokenSegue if you have multiple point in time (P585) qualifiers and multiple number of viewers/listeners (P5436) qualifiers on a single number of viewers/listeners (P5436) item, how are you reliably (without assuming larger views means a further date) supposed to know which dates correlate to which views? Lectrician1 (talk) 18:06, 16 February 2022 (UTC)[reply]
I think we're misunderstanding one another. I'm proposing you add one top level number of viewers/listeners (P5436) statement per point in time. This is how we store things like twitter follower counts over time already. I think you're suggesting using it as a qualifier. BrokenSegue (talk) 18:08, 16 February 2022 (UTC)[reply]
@BrokenSegue Because that doesn't make ontological sense. A video does not naturally have a number of likes. A YouTube video does. That is a poor way to make a relationship. Lectrician1 (talk) 18:54, 16 February 2022 (UTC)[reply]
A video does have a number of likes on a certain platform. I don't disagree that we could model it the way you want but I don't see it as required. There are a lot of ontological shortcuts that we regularly take on wikidata and this one isn't even notably bad. The cost of your ontological purity is that now anyone dealing with video items has to consider the possibility that an item representing a video will be sharded across N items each representing that video on some platform. BrokenSegue (talk) 20:24, 16 February 2022 (UTC)[reply]
@BrokenSegue Um this would be a lot of shortcuts taken if we're going to document views, etc. on possibly thousands of items. This isn't a matter of "ontological pureness". What you're suggesting is just straight out ontologically wrong. No editor or data consumer would ever want to deal with that kind of data structure.

A video item would not be "sharded" across different items. In fact, this property makes everything crystal clear. People can easily find what platforms a video is posted on and number of likes etc. for every platform. Lectrician1 (talk) 23:44, 16 February 2022 (UTC)[reply]

@Lectrician1, Ainali, Dominic, BrokenSegue, Tinker Bell: Not done: two months later, the final conclusion was not formed. —MasterRus21thCentury (talk) 19:04, 14 April 2022 (UTC)[reply]