Wikidata talk:WikiProject Video games

From Wikidata
Jump to navigation Jump to search
Home
Gamepad.svg
VG Task force, Wikidata.png

The WikiProject dedicated to the world of video games!

Current discussions


Game mode values P404[edit]

I feel like some of the values of this property are somewhat redundant and i wanted to hear other people's thoughts.

--Trade (talk) 20:55, 17 March 2022 (UTC)Reply[reply]

I agree, it's also confusing which should be a game mode versus a genre. Also, a game can be multiplayer and co-op at the same time, without having two separate game modes. Typically I just add anything that is a subclass of game mode as game mode, if it fits. AntisocialRyan (Talk) 02:41, 18 March 2022 (UTC)Reply[reply]
Your bullet points 1/ and 3/ are both about local co-op (Q107494132)?
I think local co-op (Q107494132) is fine, it’s indeed a subclass/mix of shared/splitscreen multiplayer (Q61005756) and co-op mode (Q1758804) (but not quite redundant)
not too sure about PvE, I don’t think it’s a genre nor is a mode, but more like a gameplay concept.
I would agree that two-player video game (Q5252671) is not to be used − to be honest if it weren’t for the sitelinks I would consider nominating it for deletion.
Jean-Fred (talk) 13:42, 18 March 2022 (UTC)Reply[reply]

@Jean-Frédéric: What do we do about deathmatch and team deathmatch? And similar future items? --Trade (talk) 16:51, 8 April 2022 (UTC)Reply[reply]

I've thought about this a bit (it's something I kinda want to add to vglist, so obviously Wikidata would need to support it in some fashion) and definitely agree it's a mess right now. Ideally we'd have more detailed and consistent data here, including max players, different game modes, better distinctions for split-screen games, local multiplayer, etc. Nicereddy (talk) 01:43, 5 May 2022 (UTC)Reply[reply]

Almost 50k video games[edit]

While checking some stuff about data coverage, I stumbled upon our statistics. Turns out: we're about to reach 50k video game (Q7889) items! That's already a ~9% increase compared to the start of the year (and we've already added more games so far in the past few months than during last year). So, keep up the good work!

Yet, as Jean-Frédéric often advocates for, we should keep trying to improve the quality of description of said items. So keep adding external identifiers (mix-n-match a few items everyday, in the catalog of your choice, it greatly helps!), add some statements (platforms and release dates are considered as the most important).

Hopefully, we'll reach 60k items by the end of the year. There's hope. But don't forget to have fun! It's already quite an achievement so far. --Poslovitch (talk) 20:53, 14 May 2022 (UTC)Reply[reply]

Aaaannd... 50k![edit]

That's it, we now have more than 50 000 video games described on Wikidata! See Special:Diff/1648878671. --Poslovitch (talk) 08:00, 28 May 2022 (UTC)Reply[reply]

Thanks for posting this @Poslovitch: − I would have missed it :)
I put together a quick stats/celebratory blog post, hoping to make some noise around it.
Good job everyone!
Jean-Fred (talk) 16:45, 7 June 2022 (UTC)Reply[reply]

Best external ID to import video game genres from?[edit]

genre (P136) is considered one of the core properties, but the only way I see is to add it manually. IGDB divides genre into three fields, which is not very handy: for Stellaris you need to look at "themes" to get 4X, and for Overwatch you need to combine "genre" with "players perspective" to get "first person shooter". MobyGames is even more difficult to parse, and it usually gives less information (Stellaris is just a Strategy + Real-time, which is a bit misleading). Steam have too few genres and instead relies on tags.

I see a way to fill some genres by searching for strict combination of tags, but I dunno what to do in general. Is there a database with a genre system that is comprehensive enough to be useful, but simple enough to make it possible to import it automatically? Facenapalm (talk) 17:56, 10 July 2022 (UTC)Reply[reply]

You can use the genres mapping we already have, see the last columns in Wikidata:WikiProject Video games/Statistics/Genre, as well as some mappings without property Mix’n’match catalogues.
(The more comprehensive genre ontology (and more alike to what we have) is probably Glitchwave, but we don’t link games to Glitchwave yet).
Jean-Fred (talk) 12:59, 12 July 2022 (UTC)Reply[reply]
We could potentially use PCGamingWiki's genre data (e.g. Stellaris, Half-Life 2, Elden Ring), although it's not as comprehensively-covered for all the games in their dataset and we don't have a mapping of PCGW genres <=> Wikidata genres yet. But it does seem to match the way we handle genres relatively well. Nicereddy (talk)

Wikidata:Tools/Enhance_user_interface#ExLudo.js[edit]

Wikidata user-script ExLudo.js example.png

Hello folks,

Just wanted to let you know that I wrote a small user-script, Wikidata:Tools/Enhance_user_interface#ExLudo.js (a complete riff on WikiProject Music’s ExMusica.js that enhances a bit the item pages about video game (Q7889) to display:

See on the right an example on Postal 2 (Q1974968).

Hope this is helpful! Let me know what you think :)

Jean-Fred (talk) 09:06, 19 July 2022 (UTC)Reply[reply]

This is awesome! --Poslovitch (talk) 09:23, 20 July 2022 (UTC)Reply[reply]

New feature: on game engine (Q193564) & physics engine (Q62505) item pages are now displayed the games using that engine − see eg Source (Q643572) or Havok (Q616957). Jean-Fred (talk) 08:14, 12 September 2022 (UTC)Reply[reply]

Another feature: on video game series (Q7058673) item pages are now displayed the game of that series − see eg Metal Gear (Q216655) or Final Fantasy (Q99416119) or Mass Effect (Q953242).
(It’s barebones at the moment − ultimately I would like to approximate the IGDB series pages, with the DLCs/expansions as sublists. Let‘s see :)
Jean-Fred (talk) 16:27, 23 September 2022 (UTC)Reply[reply]

ESRB Ratings and the ESRB database[edit]

ESRB rating (P852) has a constraint that requires a sister ESRB game ID (P8303) statement. However, if you've ever looked for games on the ESRB's Ratings database, you'll know that a ton of games are just completely missing from the ESRB's official database. As such, they have no known ESRB ID. For example, Bear & Breakfast (Q113255270) has an eShop page which lists an ESRB rating of E10+, but the game is not in the ESRB Ratings database.

Should we remove that constraint, and if so are there specific sources we believe should be considered trustworthy for importing ESRB Ratings? e.g. Steam, GOG.com, Nintendo eShop, etc? Nicereddy (talk) 18:34, 24 July 2022 (UTC)Reply[reply]

I was about to agree and move it as suggestion constraint (Q62026391) ; but actually we can just add a P8303 with 'no-value'?
As for sources − I see that @Trade: had started a table at Property_talk:P852#Possible_sources.
Jean-Fred (talk) 19:43, 24 July 2022 (UTC)Reply[reply]
We could use novalue, I suppose 🤔 Nicereddy (talk) 19:50, 24 July 2022 (UTC)Reply[reply]

Zombie as main subject[edit]

Is there any reason to use main subject > zombie video game (Q113259205) when genre > zombie apocalyptic fiction (Q2633346) exists? Trade (talk) 21:06, 24 July 2022 (UTC)Reply[reply]

I see you added a constraint already. I don't think that's a good idea because there are many non-video-game items about zombies. Also, there are only 24 video games with zombie (Q9406) or zombie apocalyptic fiction (Q2633346) as main subject, so cleaning up manually is very feasible. Just for fun, here's a query for video game subjects by count:
SELECT ?subject ?subjectLabel ?count WHERE {
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  {
    SELECT ?subject (COUNT(?item) as ?count) WHERE {
      ?item p:P31 ?statement0.
      ?statement0 (ps:P31/(wdt:P279*)) wd:Q7889.
      ?item wdt:P921 ?subject.
    }
    GROUP BY ?subject
  }
}
ORDER BY DESC(?count)
Run it!
Dexxor (talk) 04:42, 25 July 2022 (UTC)Reply[reply]

arcade games conflict[edit]

See Wikidata:Interwiki conflicts/Unresolved/2022#arcade_game and Wikidata:Project chat/Archive/2022/02#Arcade video games and arcade games. --Matěj Suchánek (talk) 07:29, 27 July 2022 (UTC)Reply[reply]

speedrun.com IDs being duplicated/case sensitivity problem[edit]

If you look at the "single value" constraint violations for speedrun.com game ID (P6783), you'll see that a significant number of them are from the ID being matched once with a partially-uppercased ID and once with a fully-lowercased ID (e.g. "Gran_Turismo_5" and "gran_turismo_5" or "Dota_2" and "dota_2"): https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations/P6783

This is partially due to the mix'n'match catalogue, which seems to sometimes pull in lowercase and sometimes pull in uppercase names? I'm not sure if the scraper changed, or something? https://mix-n-match.toolforge.org/#/catalog/2631

I'm going to write a script to fix these duplicates and have everything use the lowercased versions, but I was wondering if there were ways to prevent this from happening again later? Can we modify the Mix'n'match scraper to consider the IDs to always be lowercase? Should we update the regex for speedrun.com game ID (P6783) to only allow lowercase letters and no uppercase letters in these IDs? Does anyone know if speedrun.com actually changed their URL format at some point? I believe the lowercase versions are the "canonical" ones – even though speedrun.com's URLs don't seem to care either way – because when you click through game links on the site, it'll always take you to the lowercased URL.

Thanks, Nicereddy (talk) 14:36, 1 August 2022 (UTC)Reply[reply]

> Should we update the regex for код игры на speedrun.com (P6783) to only allow lowercase letters and no uppercase letters in these IDs?
I did exactly that with Microsoft Store product ID (P5885) (it had the same propblem). Although I haven't updated Mix'n'match catalogue because I don't know how to do that. Facenapalm (talk) 15:19, 1 August 2022 (UTC)Reply[reply]
Thanks for starting the discussion @Nicereddy: − that was on my mind for a while too.
The speedrun.com IDs are case-insensitive, but they definitely used to expose this kind of nice-looking uppercasing (`Gran_Turismo_5`) on their index pages, and that’s what the scraper used to pick up. I believe they have at some point changed the index pages to have the lowercase − and so the scraper picked up these the next time.
My plan would be:
  • Add an {{Autofix}} to change all to lowercase
  • Either fix the duplicates ourselves, or just wait it out (there’s a bot I believe which removes duplicate IDs)
  • Disable the current mix’n’match catalogue, which is broken beyond repair now with these mixed IDs
  • Make a new catalogue (I might make it on their API rather than on the HTML index page).
How does that sound? Jean-Fred (talk) 16:30, 1 August 2022 (UTC)Reply[reply]
@Jean-Frédéric: Oh neat, I didn't know about Autofix, that's really useful. This works for me, I think in the interim - until we have the replacement catalogue ready - maybe we should just mark the MnM as deprecated? Nicereddy (talk) 23:40, 1 August 2022 (UTC)Reply[reply]
Alright, I've added the autofix now, so hopefully KrBot will handle things for us. I'll update the regex now too. Nicereddy (talk) 23:47, 1 August 2022 (UTC)Reply[reply]
Yes, please mark the MnM as deprecated. Let's avoid having to run another autofix in a few months. --Poslovitch (talk) 13:04, 2 August 2022 (UTC)Reply[reply]
Done. I have made a new one at mixnmatch:5346, but not sure yet it’s working :/ Jean-Fred (talk) 13:24, 2 August 2022 (UTC)Reply[reply]
@Jean-Frédéric: it looks like it's working to me! I might purge the automatches and use "automatch from other catalogs" later if you're alright with that, since I found that that ended up with better results on the older speedrun.com MnM catalogue. Nicereddy (talk) 23:30, 2 August 2022 (UTC)Reply[reply]
Yeah, it was not working when I had iterate over the API by batches of 1000 or 100, but somehow it worked with batches of 20 ¯\_(ツ)_/¯
Also ran several MnM manual syncs − I’m wondering about these “673 connections on Wikidata, but not on MnM” − let’s wait for the catalog report.
Re:purging automatches: sure, feel free :)
Jean-Fred (talk) 06:52, 3 August 2022 (UTC)Reply[reply]

Video game vs Video game project[edit]

For a lot of games that were announced at Gamescom yesterday, the games were defined as instances of video game project (Q64170203) rather than video game (Q7889). I'd been changing them back to the "video game", but I was wondering if there was a consensus on this? I get that there's a distinction that might be useful for games that aren't released yet, but it seems overkill to me, and the games will end up with an "outdated" instance of value once they actually reach their release date. I don't know if the distinction is particularly worth the trouble, honestly.

Do people have opinions on whether we should distinguish between video games that have been released and those that haven't in this way?

cc: @Xezbeth since you seem to be the main editor pushing for this.

Thanks, Nicereddy (talk) 14:25, 24 August 2022 (UTC)Reply[reply]

I use it because it exists, I don't actually care either way. I assume it was created to perform the same function as film project (Q18011172)Xezbeth (talk) 14:26, 24 August 2022 (UTC)Reply[reply]
I have the same feelings as Nicereddy, but I would add instance of (P31) video game (Q7889) statements to video game project (Q64170203) items instead of replacing the instance of (P31) value—this way, we don't loose any information. Ideally, it should be possible to query/detect all video games on Wikidata using instance of (P31) video game (Q7889), but there are 850 Wikidata items which are instances of free or open-source video game (Q21125433), video game reboot (Q111223304), mobile game (Q1121542), browser game (Q848991), or similar. Dexxor (talk) 08:31, 25 August 2022 (UTC)Reply[reply]

Translation vs transliteration[edit]

Let's say there's Russian game titled in Russian that was never released in EU or AM, therefore it was never translated into English. What should we put into English label? We can take Nuclear Titbit (Q4536306) as an example, but I'm interested in general solution.

  • We can keep original Russian spelling (RAWG does exactly that, and for some reason the same did IGDB). That way the item would be clearly identified and it might be easier to search for sources, but most of Wikidata users wouldn't be able to type, read or understand this label.
  • We can translate the title into English (UVL did that). This way it would be easier to read and understand the label, but I doubt it would be any recognizable.
  • We can transliterate it, type it with Latin alphabet trying to keep original pronunciation. Main problem is, there's usually lots of ways to transliterate a single word. "Ядерный" in different sources was transliterated as Yadernyj, Jadernyjj, Yadernyy, Jadernyj. I doubt that choosing any of these variations would help users to identify the item, and it still hard to read and understand.

Considering that there's no consensus among external sources (or there's no non-Russian sources at all), what solution we should choose by default? Facenapalm (talk) 16:08, 24 August 2022 (UTC)Reply[reply]

Interesting question. I guess this applies to more than games though ; I wonder if other projects (albums, books, movies etc.) have come up with a solution already? Jean-Fred (talk) 16:30, 24 August 2022 (UTC)Reply[reply]
I think albums is slightly different story. You don't need to know the language of the lyrics to enjoy the music, so the vast majority of albums are being officially sold/streamed/etc without any translation, unlike video games. So transliteration is the way to go (Zvezda po imeni Solntse (Q1648395), Goi, Rode, Goi! (Q1534398)). Not the original spelling tho, I guess it's worth noting.
When it comes to books and films, I don't know the solution either. Facenapalm (talk) 17:00, 24 August 2022 (UTC)Reply[reply]
We could ask in the Project chat to see if anyone is familiar with how books/movies handle this. Nicereddy (talk) 23:33, 24 August 2022 (UTC)Reply[reply]

ESRB game ID mix'n'match catalogue[edit]

I wrote a script to pull down every game in the ESRB Ratings database and then uploaded the CSV to mix'n'match. We can now match ESRB game ID (P8303) to video game items on Wikidata much more easily :) The mix'n'match catalogue is here: https://mix-n-match.toolforge.org/#/catalog/5412

It's currently running an automatch job, but there are plenty of preliminary matches to evaluate already, so please have a go at it!

I'm going to write some scripts to pull in the relevant ESRB ID qualifiers and ESRB Rating automatically, so don't worry about manually filling those out. The only manual work we need is getting the ESRB game ID.

Thanks, Nicereddy (talk) 16:11, 26 August 2022 (UTC)Reply[reply]

Cool, thanks! Jean-Fred (talk) 19:20, 26 August 2022 (UTC)Reply[reply]
Thanks ! --Poslovitch (talk) 13:47, 27 August 2022 (UTC)Reply[reply]

arcade game (video and non-video)[edit]

Hello folks,

The items regarding “arcade games” are a bit of a mess, and there was some repurposing happening too recently. I have tried to clean up a bit (based on the what the items were originally, and what their usage is). We now have:

I tried to make the English labels a bit more descriptive (all of these could be “arcade game” otherwise) hence the “arcade-style video game” (take that one from the OLAC external ID)

The sitelinks are quite mixed up, and I would appreciate some help moving them around (plenty of languages I don’t speak at all in there ^_^)

We might also need to clean-up the usage of these − I corrected all the platform (P400) arcade-style video game (Q15613992) ; but there’s quite a lot of genre (P136) arcade video game machine (Q192851) to fix.

Overall, I think we have open questions on how to model arcade games (typically, do we use - platform (P400) arcade video game machine (Q192851) or the more precise platform (P400) Namco System 12 (Q442392)?) − I am not particularly knowledgeable on arcade games so I have not touched that aspect too much so far :)

Jean-Fred (talk) 12:36, 5 September 2022 (UTC)Reply[reply]

(Also relevant is video game arcade cabinet (Q1349717)) Jean-Fred (talk) 12:38, 5 September 2022 (UTC)Reply[reply]

Semi-automatic import of Nintendo64Ever ID[edit]

Hello. I was looking for a fun project to do, and just thought "why not make some data matching with OpenRefine (Q5583871)". So I did it, but was unsure if that required any special permission with bot and co. I asked on the general discussion, and I was said it would be a good idea to ask at the appropriate WikiProject. I plan on just doing the Nintendo64EVER ID (P10137) import first, but the database also have nice additional entry, like avalaible language, developer and publisher, barcode, game id (but it seems there are no property for this) and publication date (all of this by revision).

This data was collected via web-scraping. Marius851000 (talk) 15:44, 5 September 2022 (UTC)Reply[reply]

I eventually did the import itself, https://quickstatements.toolforge.org/#/batch/99058 Marius851000 (talk) 17:27, 23 September 2022 (UTC)Reply[reply]

@Marius851000: Apologies for the lack of answer!
Thanks for the matching! I have synced the mix’n’match catalogue mixnmatch:4631 − it highlighted two bad mixnmatch matches (the 2010 GoldenEye ; Jeopardy! (Q110041222) − both fixed) and a likely duplicate Rally '99 (Q3418138) vs Rally Challenge 2000 (Q518629). Thanks again!
Jean-Fred (talk) 19:55, 23 September 2022 (UTC)Reply[reply]

Handling of virtual reality video games[edit]

I was creating the Wikidata item for the upcoming "sequel" (it doesn't appear to be called an actual sequel anywhere by the developers, so I haven't marked it as such) to Boneworks, BONELAB (Q114172031) and had a few questions that I'm not sure if we've answered previously (I feel like I've asked about VR before but I don't know if it included answers to these questions?).

First, what is the "canonical" way to differentiate "normal" video games from virtual reality-exclusive video games. Are we supposed to determine this based on input method (P479)? Or I guess output method (P5196)? Should we set the instance of (P31) to "virtual reality video game"? Using "instance of" seems like a better solution for distinguishing games that are VR-only from non-VR games. Of course, there are also games that allow VR play but don't require it, like No Man’s Sky (Q15297043). So handling those also needs to be considered.

Second, how do we handle standalone headsets like the Quest and Quest 2 as supported platforms? This game is releasing on Windows and Quest 2, but I can't set Quest 2 as the platform (P400) without triggering a constraint violation. The Quest line is technically a platform itself, so it seems like we should allow it as such? Similar to how we differentiate platforms between PlayStation 3, 4, and 5.

And then there are other things to consider. For example, should we track the "play area" options? The Steam page for the game lists Bonelab as supporting Sitting, Standing, and Room-Scale VR play.

Thanks, Nicereddy (talk) 17:07, 24 September 2022 (UTC)Reply[reply]

For lack of a better way, how about BONELAB (Q114172031) has quality (P1552) VR video game (Q87741364)?
I would disagree with the use of instance of (P31) − I don’t think VR games are any more 'special' than, say, handheld video game (Q5647425) or PC game (Q4485157) which we also do not use (I’m a strong believer of only ever using video game (Q7889) as P31).
Not too sure about the platform, would need to think more about it :)
Jean-Fred (talk) 07:18, 25 September 2022 (UTC)Reply[reply]
Just use both video game (Q7889) and VR video game (Q87741364) as values for P31. has quality (P1552) is already used in many different contexts. Dexxor (talk) 10:07, 25 September 2022 (UTC)Reply[reply]
What do you mean? Per Wikidata:WikiProject Video games/Statistics/has quality, has quality (P1552) is barely ever used on Q7889 items. Jean-Fred (talk) 13:38, 25 September 2022 (UTC)Reply[reply]
I mean that has quality (P1552) is used across fields for different purposes, see random uses of "has quality". Many of these uses don't align with the property examples on has quality (P1552). On top of that, the property is also used as qualifier on Twitter username (P2002) and Discord server numeric ID. With the usage of "has quality" being such a mess, you could argue that it doesn't matter if we add yet another purpose to it, but I feel that instance of (P31) VR video game (Q87741364) is a more natural relation than has quality (P1552) VR video game (Q87741364). —Dexxor (talk) 06:46, 26 September 2022 (UTC)Reply[reply]
That’s true, but I don’t think that’s necessarily an issue. I see the meaning of properties as contextualised to the type of entities it’s used on. In this case I took my cues from WikiProject Music − who uses "has quality" to express characteristics that don’t really fit in genre − see Wikidata:WikiProject_Music/Statistics/Albums_by_characteristic. But then again I feel very strongly on a single P31 :) Jean-Fred (talk) 08:55, 28 September 2022 (UTC)Reply[reply]

How to model developers and publishers[edit]

FYI, I have raised the topic on how to model developers and publishers at Wikidata_talk:WikiProject_Companies#Businesses_and_role. Jean-Fred (talk) 08:37, 28 September 2022 (UTC)Reply[reply]