BorkedBot

Would it be possibly to have your bot update YouTube channel ID (P2397) as well? @BrokenSegue:--Trade (talk) 20:13, 26 September 2020 (UTC)

@Trade: update how? The bot is currently only approved to do a few things related to goodreads identifiers. I was planning on having it do related tasks for other identifiers (twitter and, yeah, maybe youtube) soon though. BrokenSegue (talk)

What qualifiers are you planning on adding to the twitter and YouTube identifiers? Do you have a model item in mind?--Trade (talk) 20:27, 26 September 2020 (UTC)

@Trade:For both I was gonna copy what @OsamaK: has been doing in QS and make it look as in Barack Obama (Q76) or Terry Crews (Q271464) though probably with subscriber count for youtube. BrokenSegue (talk) 20:34, 26 September 2020 (UTC)

How about copying Cliff Harris (Q70896802) but with 'named as' instead of 'website username' (Youtube channel)? I like that one the best--Trade (talk) 20:41, 26 September 2020 (UTC)

Sounds good to me but it will depend a little on what the Youtube API gives me and what they limit me at. BrokenSegue (talk) 21:13, 26 September 2020 (UTC)

So what other identifiers does this bot support other than YouTube? @BrokenSegue:--Trade (talk) 19:28, 2 December 2020 (UTC)

@Trade: it supports goodreads author/work/book and once I finish doing youtube (will take a week probably) I will be doing twitter accounts. I haven't yet set the bot up to run automatically/periodically (once a quarter or so). That is all the bot is approved to do though I have an open request to deal with stock tickers. BrokenSegue (talk)

What about Youtube channel ID's that already have some qualifier? Does the bot work on them?--Trade (talk) 22:12, 3 December 2020 (UTC)

@Trade: yes. this was demonstrated in the application for bot approval. currently it's prioritizing ones with less data but not exclusively see for example this edit. BrokenSegue (talk) 22:19, 3 December 2020 (UTC)

Can the bot also add has characteristic (P1552) items such as YouTube Vevo channel (Q72108022), Official Artist Channel (Q72112388), YouTube auto-generated artist channel (Q72108010), YouTube auto-generated topic channel (Q61948876), YouTube auto-generated artist channel (Q72108010) and YouTube auto-generated video (Q64727712) to YouTube channels?--Trade (talk) 09:01, 4 December 2020 (UTC)

@Trade: yeah I have a todo to try to handle those cases. unfortunately the youtube api doesn't return that kind of information in a structured manner and it's unclear how to reliably parse it out from what they give me. if you know how to fetch or infer it reliably I'd do it. BrokenSegue (talk) 17:22, 4 December 2020 (UTC)

verified account or profile (Q28378282) YouTube channels have BADGE_STYLE_TYPE_VERIFIED in their source code. Is that helpful? @BrokenSegue:--Trade (talk) 10:19, 11 December 2020 (UTC)

@Trade: yes but I am using their API which doesn't return that. I could scrape their website but that is against their TOS and could get me blocked. Using their API is the "correct" way. Anyone else is free to scrape their webpages or maybe I will decide it's worth it someday. There's a lot more legit data to import first. BrokenSegue (talk) 16:47, 11 December 2020 (UTC)e y

Have you considered using a query to list all YouTube channel with ' - topic' in their user name (named as)? This should help you find the autogenerated channels. --Trade (talk) 11:00, 17 December 2020 (UTC)

Yes, I considered this. I wasn't sure what the false positive rate would be. BrokenSegue (talk) 16:46, 17 December 2020 (UTC)

Limiting the query to humans and musical ensembles (+subclasses) should get rid of the false positives. AFAIK autogenerated artist channels are the only ones used on humans. --Trade (talk) 10:06, 18 December 2020 (UTC)

P2969

This edit is an incorrect use of Goodreads version/edition ID (P2969). That property should only be added to editions of books, not to the data item for the creative work itself. The properties Goodreads version/edition ID (P2969) and Goodreads work ID (P8383) should never appear on the same data item because they are used for different kinds of entities. --EncycloPetey (talk) 00:31, 15 October 2020 (UTC)

@EncycloPetey: that edit just added subject named as (P1810) to an existing entry use of Goodreads version/edition ID (P2969). That property was actually added in this edit by someone else. BrokenSegue (talk) 00:59, 15 October 2020 (UTC)

It might be better to use the bot to first remove the incorrect usages of the property before adding data to them. Adding data to misused properties will gain nothing, since it will be removed anyway. --EncycloPetey (talk) 15:18, 15 October 2020 (UTC)

@EncycloPetey: I will not remove data with a bot unless there is a lot of confidence. maybe the item's instance of (P31) is wrong. I also disagree that this is a misuse. BrokenSegue (talk) 17:39, 15 October 2020 (UTC)

The property's description of use says that it's a misuse, and that it should not appear on the same data item as the other property. --EncycloPetey (talk) 17:53, 15 October 2020 (UTC)

Yes, I disagree with the property's description. BrokenSegue (talk) 18:06, 15 October 2020 (UTC)

If you disagree with the property's description, you may work for its change but you cannot simply violate it. As for the confidence, it should be done the opposite way then you suggest: confidence is needed before the data is added and also to keep the data, if there there is a lack of confidence, the data must be removed. --Jan Kameníček (talk) 20:53, 19 October 2020 (UTC)

The data I'm adding is not incorrect. It is true that the work associated with this book version/edition is what I indicated. The state of books in wikidata is such that works/versions are conflated basically everywhere. Resolving that is outside the scope of what I was doing. For what it's worth goodreads work IDs are close to useless anyways as goodreads exposes no API for fetching by work ID. BrokenSegue (talk) 02:32, 20 October 2020 (UTC)

Danish: An essential grammar, 1st edition (Q74657280) is an example where the bot fails adding both an edition and a work link. The instance of (P31) is set to version, edition or translation (Q3331189) so the bot should not add a work-ID. — Finn Årup Nielsen (fnielsen) (talk) 10:50, 18 October 2020 (UTC)

@Fnielsen: that item is clearly already broken as it has two ISBNs which doesn't make sense for a single version/edition. BrokenSegue (talk) 14:53, 18 October 2020 (UTC)

Bohemia's case for independence is another example of the property's wrong usage. 1) you cannot add a work ID to an item of an edition, 2) you cannot link the 1971 edition in Goodreads database with WD item of the 1917 edition. I have removed these changes at this particular item, but if you did the same with other works and editions, it should definitely all be reverted. --Jan Kameníček (talk) 20:53, 19 October 2020 (UTC)

Now I have noticed that I am also partly responsible for the chaos with this particular item, as I added wrong ISBN to the edition long time ago, which probably caused the mistake with adding the Goodreads ID of the 1971 edition by your bot (although for a bot it should not be difficult to notice such discrepancies and add data only in clear cases, which was neither this one nor the one described above by Fnielsen). So I apologize for part 2) of my message above, but I still stand behind my statement in part 1) that work IDs cannot be added to editions. --Jan Kameníček (talk) 22:47, 19 October 2020 (UTC)

Actually noticing these discrepancies is non-trivial I'm afraid. Often the goodreads title is very different from the wikidata one (even in a different language). I do do some filtering when matching authors to goodreads author IDs though. The reason I added subject named as (P1810) was to make catching these errors easier though in this case... I argue removing the link to the goodreads work from that item did more harm than good since there is no existing item for the work itself. BrokenSegue (talk) 02:32, 20 October 2020 (UTC)

0 works and 0 views for auto-generated youtube channels

Hi, here you have added few zeros for autogenerated channel. This may confuse end user, because this channel is not empty, topic-related content is distributed between playlists (I don't even think that Youtube can create auto-generated empty channel). Maybe it would be better to just add qualifier has characteristic (P1552) YouTube auto-generated topic channel (Q61948876) instead of 0/0? --Lockal (talk) 08:00, 18 December 2020 (UTC)

@Lockal: Somehow I missed this message. Hmm, yeah I'm just reporting what the youtube API gives me. I agree it's confusing and optimally we just wouldn't link "topic" channels at all. I think I'll eventually go back and mark them all as autogenerated topic channels at some point if someone doesn't beat me to it. BrokenSegue (talk) 16:25, 24 December 2020 (UTC)

'There's a lot more legit data to import first.' That sounds interesting. Want to tell us about it? --Trade (talk) 10:43, 24 December 2020 (UTC)

@Trade: for example Twitter user numeric ids. BrokenSegue (talk) 16:07, 24 December 2020 (UTC)

Would you consider taking a look at importing review scores from IMDB? See an example here tt9876344 > Twinkle Nora Rock Me!? --Trade (talk) 20:32, 25 December 2020 (UTC)

@Trade: I would consider it but I don't think IMDB has a free public API and I don't think they'd appreciate this data being reposted even if I did pay to get access (they might just straight up deny me). Some brief googling suggests maybe >$120/yr would get me access but I don't know what the terms of use would be (maybe I can get Wikimedia to foot the bill). Scraping is an option but again they won't look kindly on that and I'd rather focus on totally above board things for now. There are other similar review score databases that are more open (e.g. Letterboxd (Q18709181)). BrokenSegue (talk) 22:47, 25 December 2020 (UTC)

Maybe try and email AniDB? They seem very open if @Shishma: is to believe. If anything they might be grateful for the free advertising--Trade (talk) 23:25, 25 December 2020 (UTC)

@Shisma:--Trade (talk) 23:27, 25 December 2020 (UTC)

If you can link me documentation for an API that I can use and that allows this use I will basically gladly import any data for you. That said I'll have to prioritize it against other imports and get approval on WD:BOT (which seems to take like months). BrokenSegue (talk) 23:39, 25 December 2020 (UTC)

@Trade, BrokenSegue:. I don't know. apparently the API allows access to ratings. You'll have to apply to be a client. The data is good, so i think it's worth a try--Shisma (talk) 10:18, 27 December 2020 (UTC)

Rank

@BrokenSegue: Now an item has multiple social media followers statement. You need to mark the most recent as preferred.--GZWDer (talk) 11:58, 15 January 2021 (UTC)

@GZWDer: oh dear you are right. I meant to do that. I'll fix it presently. BrokenSegue (talk) 14:31, 15 January 2021 (UTC)

YouTube video ID

Is there a reason why you use "named as" instead of "title" as qualifier?--Trade (talk) 21:16, 17 March 2021 (UTC)

@Trade: not really. I can change it? honestly not sure when to use which. it is both named as and titled that and the property permits both. BrokenSegue (talk) 01:36, 18 March 2021 (UTC)

Title would be preferable considering that the videos itself are works seperate from the subject itself. --Trade (talk) 08:24, 18 March 2021 (UTC)

@Trade: ah I think I know why I didn't do it that way. I went to change it and remembered that title requires me to annotate the language of the title while "named as" doesn't. The API doesn't give me the language of the title just (sometimes) the language of the video. There's no obvious solution there without just defaulting to english or something. BrokenSegue (talk) 14:13, 18 March 2021 (UTC)

I think a good solution would be to prevent the bot from adding 'named as' to Youtube videos that already have 'title' as a qualifier. --Trade (talk) 15:08, 21 March 2021 (UTC)

@Trade: ok I have done that. BrokenSegue (talk) 01:53, 23 March 2021 (UTC)

I have never seen a YouTube video listing the language of the video. Or is that only available in the API? --Trade (talk) 14:26, 18 March 2021 (UTC)

Also, i can see that duration and number of views are missing? Was there a problem getting those? --Trade (talk) 14:27, 18 March 2021 (UTC)

@Trade: yeah the API returns some language codes. Unclear how accurate it is though and it isn't always populated. Number of views is being populated e.g. [1]. Duration was a little tricky and I decided to do it in a second pass. Though maybe I'll stop it and try sooner. The main limit in getting this done is the slowness in writing to wikidata so a second pass adding a single property doesn't slow the process down much. BrokenSegue (talk) 15:38, 18 March 2021 (UTC)

What about items whwre the YouTube video is the subject? --Trade (talk) 15:53, 18 March 2021 (UTC)

What about that? I'm not following. How do I even detect that. BrokenSegue (talk) 16:27, 18 March 2021 (UTC)

I don't think there's a way to detect that other than making a list manually. I apologize, that was an unreasonable request of me to make. --Trade (talk) 15:03, 21 March 2021 (UTC)

Uncertainties on times

Re [2], why? Better to just say 1716 seconds and leave it at that, rather than adding ±1. Thanks. Mike Peel (talk) 18:14, 24 March 2021 (UTC)

@Mike Peel: the videos have sub-second durations but the API rounds the seconds off. So the duration is plus or minus one second. BrokenSegue (talk) 18:36, 24 March 2021 (UTC)

@BrokenSegue: Just round it to the second, then. There's no need for the ±1 if it's never going to be ±2 etc., as that's implied. Otherwise, where do you stop - at a millisecond, nanosecond, ...? Thanks. Mike Peel (talk) 19:06, 24 March 2021 (UTC)

@Mike Peel: If you look the times I'm putting often do not match the times listed on the youtube video player because of this rounding. I would prefer to have the exact length (i.e. number of frames * frames per second) but I do not. But there is an exact answer and there is a place to stop if someone else wanted to use a different API to get the actual precise duration. There is no downside to listing the data as I am and doing so is more correct. Plus or minus 1 is not implied in all second length measurements. Why would I stop. BrokenSegue (talk) 19:41, 24 March 2021 (UTC)

@BrokenSegue: Then ... fix it? Use a different interface if the API is being limiting, or store the values differently. But ±1 does not make sense. Happy to escalate if needed... Thanks. Mike Peel (talk) 19:45, 24 March 2021 (UTC)

@Mike Peel: escalate to google? The official youtube API only provides this resolution. If you were to scrape the youtube webpages you could get finer resolution but google would ban me. ±1 makes sense. BrokenSegue (talk) 19:47, 24 March 2021 (UTC)

@BrokenSegue: Other admins to potentially block the bot. There's a long history here with ±1, it was originally added by default by Wikibase for anything with a quantity, but was subsequently removed. If you want to stop the bot and discuss it, that would be OK with me, but please don't continue without fixing it. Thanks. Mike Peel (talk) 20:02, 24 March 2021 (UTC)

@BrokenSegue: Also, I've just looked and can't find the bot approval for this task, please can you link me to it? Thanks. Mike Peel (talk) 20:07, 24 March 2021 (UTC)

@Mike Peel: feel free to try to get this bot banned. it was approved and your complaint is trivial and without merit. you can find the approval at Wikidata:Requests_for_permissions/Bot/BorkedBot_4. Have fun. BrokenSegue (talk) 20:33, 24 March 2021 (UTC)

@BrokenSegue: What's with your attitude here? Whenever someone raises an issue with a bot I operate, I take it seriously. Anyhow, now raised at Wikidata:Administrators'_noticeboard#Possible_issue_with_User:BorkedBot_-_please_can_someone_check%3. Thanks. Mike Peel (talk) 20:40, 24 March 2021 (UTC)

@Mike Peel: If you look at my talk page and elsewhere on this page you will see I usually take them seriously too. This complaint is trivial. BrokenSegue (talk) 20:42, 24 March 2021 (UTC)

@BrokenSegue: So it's trivially fixed? Thanks. Mike Peel (talk) 20:48, 24 March 2021 (UTC)

@Mike Peel: yes it would be a trivial change. but it would be a change for the worse. BrokenSegue (talk) 20:50, 24 March 2021 (UTC)

@BrokenSegue: I disagree. But let's see what others think. No hard feelings either way. Thanks. Mike Peel (talk) 21:16, 24 March 2021 (UTC)

Other social media

Are there any plans for the bot to perform similar functions as it does on YouTube and Twitter for other social media, like Instagram or TikTok? If not, do you know if anyone else is currently working on something that would cover those platforms? DoSazunielle (talk) 00:17, 11 April 2021 (UTC)

@DoSazunielle: I am willing to do this but I am unsure if there are open APIs I can use to do this. I do not think there is one for instagram. I haven't looked into tiktok yet (I don't know how many tiktok accounts we even have yet). BrokenSegue (talk) 02:58, 11 April 2021 (UTC)

If it seems like TikTok will work after you look into it, let me know. I spend time populating social media accounts for many items and would be willing to focus on more TikTok handles (as well as important TikTok users and accounts whose pages are yet to be created) if this ends up happening. DoSazunielle (talk) 03:03, 11 April 2021 (UTC)

@DoSazunielle: Looking at https://developers.tiktok.com/ doesn't make me think they have an official API that would let this work. It seems there's a reverse engineered API some people are using but using that wouldn't be very high on my priority list because I fear being blocked. This is especially true because we only have 2k tiktok usernames (compared to like 200k+ for yt and twitter). I'm not saying I'll never do it but there's other things higher on my priority list. I'd probably do instagram using an unofficial API before I'd do tiktok though just because of data coverage. BrokenSegue (talk) 03:46, 11 April 2021 (UTC)

Also add MediaWiki page ID (P9675) qualifier to Fandom article ID (P6262)

this bot does some great work, adding both the qualifiers subject named as (P1810) and language of work or name (P407) to all Fandom article ID (P6262) statements. It could be helpful to also have the MediaWiki page ID (P9675). It can be easily extracted from every MediaWiki article using the regular expression /"wgArticleId":(\d+)/. This property would allow us to track whether a mediawiki article has been moved to a different lemma.

I have drafted this here. Thanks :) --Shisma (talk) 17:34, 30 July 2021 (UTC)

@Shisma: Thanks, I'm not sure I understand though. You want the cur id of the fandom page at the current time? Don't the curids of pages change as people edit them or am I confusing this with revid? Also, I'm not sure what I should be running that regex against? The html of the mediawiki article? Overall though this seems sensible. BrokenSegue (talk) 20:56, 30 July 2021 (UTC)

Should this also be done with Super Mario Wiki ID (P8256)? DogsRNice (talk) 02:10, 14 April 2022 (UTC)

it should be done for all mediawiki external links. BrokenSegue (talk) 00:34, 16 April 2022 (UTC)

Yeah that would make more sense DogsRNice (talk) 21:35, 27 April 2022 (UTC)

@BrokenSegue: the name is somewhat confusing. Where the oldid references a particular revision of an article, whereas curid references the current state of the article. Unlike the lastrevid which is the revision which is currently the latest revision. You'll be always able to find this talk page under wikidata.org/wiki/?curid=97564323 unless to delete it.

The regex was meant to be run against the html, but I just noticed, you can also acquire the id from the api: ?action=query&prop=info&titles=$lemma (named here as pageid). This property needs a lot of aliases 😅--Shisma (talk) 10:58, 31 July 2021 (UTC)

ok cool thanks for the API tip. much easier for me. I updated the code to set curids. See this example. I won't go back and update all the old entries at least for a while as I have a backlog of large tasks to do. Ping me if I never get around to it. BrokenSegue (talk) 13:57, 31 July 2021 (UTC)

👌 Shisma (talk) 17:22, 1 August 2021 (UTC)

Converting P6742 to it's numerical identifier

Would this be possible for your bot to do? --Trade (talk) 21:55, 14 August 2021 (UTC)

@Trade: I don't quite understand. It already looks numerical? Is there some other identifier or property I should look at? Looking at their API seems like I would only get 2k requests per day so any work would be slow. BrokenSegue (talk) 15:35, 15 August 2021 (UTC)

Musixmatch uses both numerical identifier and URL slug. I am asking if it's possible for the bot to convert items that ues the URL slug (such as Qu'est-ce qu'on attend (Q105744697)) to their respective numerical identifier. --Trade (talk) 17:39, 15 August 2021 (UTC)

@Trade: is there info on how to go from the slug to the number for Musixmatch track ID (P6742)? Also, I'm unsure a bot is needed given the identifier is used less than 100 times. BrokenSegue (talk) 18:36, 15 August 2021 (UTC)

The reason why the property is used so little is because the numerical identifier makes it so cumbersome. If we could use the slug and lwt the bot convert it, more people would use it--Trade (talk) 18:43, 15 August 2021 (UTC)

I see. Maybe we should do like what we do for other properties and make a slug property and have a bot populate the numeric one. BrokenSegue (talk) 19:06, 15 August 2021 (UTC)

"Named as" for MediaWiki pages

Hi, please, could you use wgTitle or og:title for Fandom and similar external identifiers? There should be no underscores or uri-encoded entities in names[3]. --08:53, 20 September 2021 (UTC)

@Lockal: good point I'll look into this. BrokenSegue (talk) 14:29, 20 September 2021 (UTC)

I looked into it and I'm very confused why it made this error. It's already looking at the title property returned from the json endpoint. I reran the bot over that particular item and it got the title correct. I'm at a loss for how it could've made this error. BrokenSegue (talk) 16:50, 20 September 2021 (UTC)

Social media followers

Hello, thanks for your bot. Do you have an idea why it hasn't added social media followers (P8687) to items such as Jiří Navrátil (Q100825776) yet? I was planning to do a little visualisation of social media followers of our parliament members :). Vojtěch Dostál (talk) 06:55, 27 October 2021 (UTC)

@Vojtěch Dostál: yeah the bot doesn't populate follower count data below some threshold. that person only seems to have a few hundred followers and the limit I think is around 5k. BrokenSegue (talk) 14:31, 27 October 2021 (UTC)

Ah, OK :) makes sense. Vojtěch Dostál (talk) 14:58, 27 October 2021 (UTC)

Steam application ID (P1733)

Would it be possible to make BorkedBot import review scores from Steam? Q107178084#P444 as an example --Trade (talk) 01:12, 23 November 2021 (UTC)

@Trade: I've looked at this and I don't think the steam API exposes this information for free. If you can find an API that returns this data I'll implement this. BrokenSegue (talk) 02:41, 23 November 2021 (UTC)

@BrokenSegue: It's public, just check https://store.steampowered.com/appreviews/271590?json=1 Germartin1 (talk) 19:25, 23 November 2021 (UTC)

@Germartin1: @Trade: I started working on a bot to import this but I don't think that link provides accurate data. Compare [4] to [5]. The number of reviews and total scores don't match. The steampowered link shows thousands of reviews and your link shows just a few hundred. The fraction of positive reviews also differs. No idea why. Also, I'm not a fan of importing review score results in the format " 71% (Mostly Positive)" (I'd prefer "71%" or "Mostly Positive") but that's just bikeshedding. BrokenSegue (talk) 03:57, 25 November 2021 (UTC)

Maybe i'm missing something but it seems the review score and number of reviews show up in the [view-source:https://store.steampowered.com/app/348710/Saint_Seiya_Soldiers_Soul/#app_reviews_hash page source]? --Trade (talk) 22:58, 20 April 2022 (UTC)

@Trade: no you're not missing anything. I've just to date been reluctant to do scraping of pages. I've preferred to use APIs. At the moment I don't have time to add more jobs to this bot. BrokenSegue (talk) 00:50, 21 April 2022 (UTC)

That's okay, you have been very productive. Could you ping me once you get time to add jobs? --Trade (talk) 01:14, 21 April 2022 (UTC)

DOB/DOD

I've seen your bot change more specific dates of birth and death to preferred rank. Is this something your bot regularly does or is this an occasional task? I've been adding dates to a large number of items assuming that a bot would rank them, and was wondering if I should stop doing this if your bot is no longer ranking dates. Please don't feel obligated to change anything, just asking so I can adjust my workflow accordingly. Thanks. Gamaliel (talk) 21:28, 7 January 2022 (UTC)

@Gamaliel: The bot does it continuously but unfortunately the code is not very optimized (due to SPARQL query timeouts) and so it takes a very long time to find/fix all the items (probably it will never get to them all). I'm hoping to rewrite it to be more robust but for the time being it is acting on a "best effort" basis. BrokenSegue (talk) 21:48, 7 January 2022 (UTC)

Thanks for the update. Appreciate your work with the bot. Gamaliel (talk) 22:14, 7 January 2022 (UTC)

Titles of recordings vs recorded events

Nice to see someone - or something, in this case - attend to my recently created item The Art of Code (Q110612096), thanks. Unfortunately the constraint validator has an issue with the qualifier subject named as (P1810) on the YouTube recording, but suggests using title (P1476) instead. And thinking about it, I kind of agree; there is nothing wrong with a recording of an event having a slightly different title than the event itself. It's not like the recording claims that the event recorded had this longer title (that includes the speaker's name), but the event organizer (NDC Conferences) themselves apparently picked the two different titles, probably to make the speaker's name stand out better in the much broader context of YouTube videos than would be necessary in their five-day conference schedule. It would be a different matter if YouTube were considered a database of live presentations (each having an original title) only, but then they would probably have a separate field for the speaker's name instead of jamming it into the title field. Your view on this? --SM5POR (talk) 07:19, 19 January 2022 (UTC)

@SM5POR: this was considered during the design of this bot. unfortunately we cannot just use title (P1476) because that requires knowing the language of the title. So the options are use subject named as (P1810), use title (P1476) but with a guessed/bad language or do neither. The first option was chosen. Do feel free to override the bot though and change it to title (P1476). BrokenSegue (talk) 15:31, 19 January 2022 (UTC)

Ok, thanks for the explanation. I see now that half of the discussion at Property talk:P1476 is about determining the language of titles with multiple languages, no discernible language etc, so I understand your problem. Yet this strikes me as another clue that subject named as (P1810) was never meant for arbitrary titles of works, just alternative names of actors, role characters and possibly authors. In that situation I would probably have left out the title entirely, leaving it for a human editor to add.

But then I see that the bot has actually added language of work or name (P407) as English (Q1860), and I wonder where it got that from? Does YouTube encode the language of the uploaded video, and how reliable is it? If there is any correlation between that piece of metadata and the actual language of the video, then I don't think it's too much of a stretch to assume that the title is written in the same language. There may be some cases where this is wrong, but I wouldn't consider it much worse than the current constraint violation. Maybe some other flag might be set saying "this property setting calls for human validation" and then have another bot patrol Wikidata for those and dispatch them to an appropriate editor (say, the one who either created the item or set the YouTube video ID (P1651) claim) for review? I'm not expecting you to do this or even change your bot accordingly, but I suggest it as a potential future development.

And since you hint that your bot won't add the subject named as (P1810) property again if I remove it and define a title (P1476) instead, I will probably do just that. Thanks. -- SM5POR (talk) 06:32, 20 January 2022 (UTC)

Preference for most precise value: January 1 (P569/P570)

Maybe the bot should skip January 1 dates. Sample: Q67202381.

I think it's more likely they are wrong than others.

Other than that, I think the bot's work is most useful. --- Jura 15:06, 1 February 2022 (UTC)

@Jura1: thanks. That's a reasonable suggestion. I'll try to implement it today but we'll see. I wish I had more time/energy to get this thing working faster. BrokenSegue (talk) 17:18, 1 February 2022 (UTC)

Thanks. No urgency though. I don't think it happens that frequently that we have both.

I'm still not really sure how to solve the January 1 problem in general (for other cases). At least, I did write it down: Help:Dates#January_1_as_date. --- Jura 17:31, 1 February 2022 (UTC)

Number of Likes, Dislikes, Comments

Would it be possible to add qualifiers to YouTube video ID (P1651) statements with these metrics? number of likes (P10649), number of dislikes (P10650), number of comments (P10651)

I also don't know if number of viewers/listeners (P5436) is added already, but that too if not. AntisocialRyan (Talk) 13:36, 18 April 2022 (UTC)

@AntisocialRyan: it would be possible but I'm kind of skeptical of adding more time sensitive data as qualifiers. we do do it but it doesn't seem right since there's no good way to indicate when the qualifiers are valid. a more practical issue is that I have much less free time to commit to wikidata for the coming months so I'm unlikely to do something unless it's high priority. BrokenSegue (talk) 23:32, 19 April 2022 (UTC)

Could i ask you to make the bot do something more simple? Mark statements with P1552 > Q105761517 as deprecated and the statements with P1552 > Q105265074 as preferred. @BrokenSegue: Kind regards --Trade (talk) 18:04, 26 May 2022 (UTC)

Names of works

Please add translated names in the appropriate language fields, or on the specific translation with that title. You are adding information to the worng location using a bot that has not been approved for that task, and which is making incorrect placement edits. --EncycloPetey (talk) 21:27, 12 May 2022 (UTC)

@EncycloPetey: if you provided more detail I would be able to determine what you are talking about. This bot does many tasks. All of which are approved. I'm going to guess you mean the fandom task? It was approved here Wikidata:Requests_for_permissions/Bot/BorkedBot_4. BrokenSegue (talk) 01:53, 13 May 2022 (UTC)

I'm talking about edits like this one, where you added an English title to a work written in ancient Greek. Your task did not not have approval to add titles to works in a language in which they were not written, and it shouldn't do that. Those are labels, and should be placed in the label field for that language (if they do not already exist there), not be added as titles. --EncycloPetey (talk) 02:22, 13 May 2022 (UTC)

@EncycloPetey: You are confused. That qualifier is about the language of the title of the article on memory beta. Which is english. The edit is perfectly fine and is exactly in line with the point of the approved bot action. I don't have time to argue but if you still feel I am mistaken please report this bot on the admin noticeboard. BrokenSegue (talk) 02:29, 13 May 2022 (UTC)

@EncycloPetey: actually I think there's just some confusion. An error was made but it was not done by my bot. Look carefully at the edit history. This is the fault of @Loominade: who was using the rather too aggressive Wikidata:Tools/Wikidata for Web extension. BrokenSegue (talk) 02:44, 13 May 2022 (UTC)

Ah, I see now. My mistake. Thanks for helping me realize what happened. --EncycloPetey (talk) 03:00, 13 May 2022 (UTC)

EncycloPetey: I thought the current consensus was that original title as well as any localised title should use the property title (P1476). But the original title should always have a preferred rank. As modelled here. This scheme was applied to the entity in this version--Loominade (talk) 08:46, 13 May 2022 (UTC)

Loominade The example you have pointed to is an Anime series, where the original video content is used with an English dub. It is not an ancient Greek epic poem. Poems are not dubbed; they are translated into a different language and published in books. Books cannot use that scheme, because translations and editions may have multiple titles in the same language. Standard library cataloging titles are placed as labels of the work data item, but each edition or translation has its own separate data item for each published edition. This must happen because each publication has unique publication data, such as the publisher, editor, translator, number of pages, place of publication, etc. The title of that edition or translation should be placed on the data item for that edition or translation, and not on the main item. --EncycloPetey (talk) 16:10, 13 May 2022 (UTC)

I see your point. I'd argue that, when somebody cites The Republic, they don't necessarily cite a specific edition of the poem but the work itself, because that's what the work is known as in English. I am not passionate about it and I am not sure if I am right. I just wanted to express that I did the edit intentionally. --Loominade (talk) 08:13, 14 May 2022 (UTC)

Your edits on Emerald Hill Zone

Uh, thank you for fixing my item, bot... 142.105.209.171 20:02, 14 May 2022 (UTC)

Why does bot add number of subscribers (P3744) and social media followers (P8687) for YouTube channel subscriber counts?

1. Why do you need to add number of subscribers (P3744) to YouTube channel ID statements AND document the subscriber count with social media followers (P8687) YouTube channel ID as a qualifier?
2. Shouldn't we be using only one property for subscriber counts? Using both shows that they are conflated and people aren't sure which to use when they are documenting subscriber counts manually...

Lectrician1 (talk) 20:26, 29 November 2022 (UTC)

@Lectrician1: we don't need to do both. This is a historical artifact. We used to not have social media followers (P8687) and so number of subscribers (P3744) was the only way to record it. When I started running the bot I kept doing it the old way for compatibility. I'd be fine with stopping doing the number of subscribers (P3744). I don't ever update its value and it's ambiguous when the time it applies to. BrokenSegue (talk) 22:08, 29 November 2022 (UTC)

That would sound good. Thank you. Lectrician1 (talk) 01:45, 30 November 2022 (UTC)

Also, we should probably clean up the old uses of number of subscribers (P3744) and move them to social media followers (P8687). Could you do that or should I make a separate bot? Lectrician1 (talk) 01:46, 30 November 2022 (UTC)

Clean up how? BrokenSegue (talk) 06:40, 30 November 2022 (UTC)

Delete the use of number of subscribers (P3744) as a qualifier on on YouTube channel ID (P2397) and move the count to a social media followers (P8687) statement. Lectrician1 (talk) 13:52, 30 November 2022 (UTC)

Please dont. Not everybody supports the social media followers property. Trade (talk) 01:12, 14 February 2023 (UTC)

YouTube channel names with special characters

Hi BrokenSegue,

I'm stumbling about the YT channel name @easylanguages (at Easy Languages Project (Q116171511)), which is getting garbled into HTML entity notation when clicked. Can your bot covert it anyways, or do I have to find the YT channel ID manually?

Best wishes-- Thelemic Magick (talk) 16:19, 11 January 2023 (UTC)

@Thelemic Magick: Hi. The problem actually isn't special characters. The problem is you are using the wrong property. Channel IDs are of the form "UCEbFsO2sM_wjTn44YAgSrrg". What you have entered is a YouTube handle. Please look at the warning next to the handle which explains that you have entered something in the wrong format. We have a separate property for youtube handles called YouTube handle (P11245). This bot does not currently convert between them but there is an open proposal to do that which you can support at Wikidata:Requests for permissions/Bot/BorkedBot 9. Also the item you made Easy Languages Project (Q116171511) needs a instance of (P31) statement to be valid. Thanks. BrokenSegue (talk) 17:59, 11 January 2023 (UTC)

Running schedule

Hello,

how often does BorkedBot run the "YouTube-ID to follower count" job? I've noticed edits on 8th and 15th of January.

Kind regards-- Antifaschistische Frontschule (talk) 17:05, 16 January 2023 (UTC)

@Antifaschistische Frontschule: roughly once per week. Sometimes it misses a week. is this a problem? BrokenSegue (talk) 17:40, 16 January 2023 (UTC)

Good to find this thread: I've just wanted to ask, whether you might do an extra-run out of your usual schedule, please?--Billy McCartney (talk) 13:49, 17 January 2023 (UTC)

@Billy McCartney: run of what task? and why does it need to be run out of schedule? BrokenSegue (talk) 17:57, 17 January 2023 (UTC)

Reference for Twitter followers

Hi! I think it'd be nice if BorkedBot, when updating Twitter follower counts, added stated in (P248)X (Q918) as a reference. Would that be possible? Cheers, {{u|Sdkb}} ^talk 20:09, 6 February 2023 (UTC)

@Sdkb: Sure I can change the code to do that going forwards but the whole task may be done because of the API policy changes. Also, I think I'd prefer something like stated in Twitter API. BrokenSegue (talk) 00:33, 7 February 2023 (UTC)

I guess that means the bot could be freed up for other tasks? Trade (talk) 01:13, 14 February 2023 (UTC)

@Trade: Freed up? The main limit on doing more tasks is on my personal time to contribute not on compute time. BrokenSegue (talk) 03:07, 14 February 2023 (UTC)

Sorry i forgot that since it's so long i have been on this user talk. Do you have time for a simple task? Trade (talk) 09:10, 14 February 2023 (UTC)

@Trade: Maybe? My time is more limited at the moment. Probably just describe the problem and I can tell you if I can do it. Or else list it on Bot Requests and ping me a link to it. BrokenSegue (talk) 22:27, 14 February 2023 (UTC)

Add "subject named as", "start time", "number of subscribers" and "point in time" to P3984. Or if that's too much the too first one are also fine. Trade (talk) 08:39, 15 February 2023 (UTC)

@Trade: It was my understanding that number of subscribers (P3744) as a qualifier was not good practice because it's impossible to update the information later. And it doesn't make sense to use social media followers (P8687) on items like Africa (Q15). I'm unfamiliar with the reddit API but I'll take a look this weekend to see what can be done. BrokenSegue (talk) 16:28, 15 February 2023 (UTC)

The way i see it: As long as identifiers such as YouTube channel ID (P2397) use number of subscribers (P3744) then other similar identifiers are allowed to do so as well. Very much appreciated--Trade (talk) 18:08, 15 February 2023 (UTC)

number of subscribers (P3744) was a terrible idea, and the longer we avoid fixing it, the more headaches it's going to create for us. See this thread. {{u|Sdkb}} ^talk 19:16, 15 February 2023 (UTC)

I mean, if I understand you correctly. your position in that thread was that all ways of representing this data that we are considering are bad. Sounds like I should make a bot to remove P3744 from P2397 so people will stop thinking it's good practice. BrokenSegue (talk) 03:59, 16 February 2023 (UTC)

Having a property for e.g. "Number of Twitter subscribers," qualified by the point in time (P585), seems like the sensible approach to me. Having the subscribers be a qualifier of the channel ID could also work alright, although then there would be no way to mark the point in time or retain historical data. It's using "number of subscribers," qualified by the platform and point in time, that I really don't like. For many entities that have lots of different social media channels, there are many different things that are all being thrust together by the ambiguity of the property (number of subscribers...of what?), which is never a sound approach to data. {{u|Sdkb}} ^talk 04:07, 16 February 2023 (UTC)

@Sdkb: The reason we didn't do that is that many items have multiple twitter accounts (for example Google). So we would be forced to qualify the follower number with the twitter account identifier anyways. At that point it makes more sense to have a generic "follower count" property. Can we not solve this problem with some Lua magic? I really would like to see this problem worked out if at all possible. I don't know Lua but I could try to learn. BrokenSegue (talk) 04:20, 16 February 2023 (UTC)

Hmm, that's complicated. I don't think too many items would have multiple accounts, but for the ones that do, we'd want the "number of Twitter subscribers" property to be for the account listed with X username (P2002).

I'm unfortunately not very familiar with Lua or how it might help here, but I'd love to hear about it! The core issue to me seems to be that we can't currently qualify qualifiers, and we'd need to in order to have account/subscriber number/time all together. {{u|Sdkb}} ^talk 06:55, 16 February 2023 (UTC)

@Sdkb: ok I made a module that fetches an item's youtube subcount using the current data scheme. here is the module en:Module:YouTubeSubscribers you can use it by including {{#invoke:YouTubeSubscribers|subCount|}} in a page (I tested this on a preview of en:PewDiePie and it worked). It's still rough around the edges but it is a proof of concept. You can also invoke it as {{#invoke:YouTubeSubscribers|subCount|qid=Q111862397}} like I did on en:Module talk:YouTubeSubscribers. If you are interested in handling the wikipedia side of this migration I can make the module more bulletproof and we can go ahead. BrokenSegue (talk) 01:11, 17 February 2023 (UTC)

@BrokenSegue, amazing!!! That is fantastic — I'd be happy to help with the Wikipedia side of things. I'll be busy this weekend, but I can jump on it this coming week or whenever once you've completed the stability improvements. Cheers, {{u|Sdkb}} ^talk 07:08, 17 February 2023 (UTC)

How'd it go? Did you got it working? Trade (talk) 22:25, 6 March 2023 (UTC)

@Trade: Sorry I totally forgot about this and didn't have time to investigate. I can try to fit it in in the next week. It sort of reminds me of this request Wikidata:Requests_for_permissions/Bot/MidleadingBot_4 which I'm iffy on. If you want you can go ahead and open the RfP for a new BorkedBot task and I can work to get the code ready in the background. BrokenSegue (talk) 22:36, 6 March 2023 (UTC)

Done Trade (talk) 01:30, 7 March 2023 (UTC)

How is the code going? Trade (talk) 01:43, 12 March 2023 (UTC)

Bot does not handle renamed Twitter accounts correctly

For bullshit legal reasons, temporaerhaus (Q27945856) have had to change their name from Verschwörhaus to Temporärhaus this month. (As you might guess from the new name, we can probably expect another rename in future, though I don’t know when it would take place.) As part of that, they renamed their Twitter account from @verschwoerhaus to @temporaerhaus this month. I updated the X username (P2002) statement accordingly yesterday, adding end time (P582) to the old statement and start time (P580) to the new one.

However, the bot has now changed the new statement, in my opinion incorrectly: it claims that the new username started back in 2016, even though that is only the date on which the account was created (under another name), not the date on which it assumed its current name. I could fix it manually, but I assume the bot will then break it again soon enough. Please advise. Lucas Werkmeister (talk) 09:58, 8 May 2023 (UTC)

@Lucas Werkmeister: unfortunately the API says that that account started in 2016. There is no way for the Bot to know that the account was renamed since it's using the same internal identifier. Fortunately, the bot (I believe) will not try to override the value again now that a twitter user id is set. So just doing nothing should work. If it breaks again I'll look into it. Also, the Twitter API has been locked down and so I don't even know why the bot is still working at all (I didn't pay Elon) so it's probably nearly moot. BrokenSegue (talk) 15:56, 8 May 2023 (UTC)

@BrokenSegue: Alright, thanks. Lucas Werkmeister (talk) 18:07, 8 May 2023 (UTC)

P11245

Your bot was approved to fix YouTube handle (P11245), but I only see the list growing on Wikidata:Database reports/Constraint violations/P11245, are you still planning to use your bot to add the missing YouTube channel ID (P2397) statements? Mbch331 (talk) 07:13, 14 May 2023 (UTC)

@Mbch331: That is still the plan but as I started to implement this I discovered that the YouTube API doesn't actually support returning this information so I need to do this another way. I am still planning on doing this but I'm not sure of the timeline. You can feel free to do it. BrokenSegue (talk) 19:32, 14 May 2023 (UTC)

The information is present in the source code of every YouTube channel page. It used to be a plain value, but that changed. Nowadays it's easiest to get the value for og:url and strip of the https://www.youtube.com part Mbch331 (talk) 06:26, 15 May 2023 (UTC)

Yeah I figured that much out. I've historically avoided doing web scraping and preferred to use official APIs for fear of getting banned. I'll eventually get around to this BrokenSegue (talk) 00:56, 16 May 2023 (UTC)

Social media followers on concepts

I'm a bit concerned about the recent edits like animal (Q729) [6] where social media followers (P8687) is added as a main statement Q729#P8687 to a concept rather than an entity (e.g. a person or group). I presume this has been done because there is a subreddit about that concept, but I don't think that really meets the meaning of "social media followers" as proposed. I think it would be better off excluded from this type of item, or added as an extra detail in a qualifier for Q729#P3984. If you intend to continue the current method, I think it would help to have a wider discussion first. --99of9 (talk) 00:46, 24 May 2023 (UTC)

@99of9: you know I had the exact same thought. Then I saw Wikidata:Requests_for_permissions/Bot/MidleadingBot_4 and I ... don't know. I've paused the bot for the time being but this issue isn't limited to this bot. BrokenSegue (talk) 06:16, 25 May 2023 (UTC)

@BrokenSegue, Midleading, Ymblanter: Thanks. We're already in a messy situation then! Where is the best forum to discuss this? --99of9 (talk) 23:07, 25 May 2023 (UTC)

I guess Wikidata:Project chat Ymblanter (talk) 05:32, 26 May 2023 (UTC)

Subreddits

A lot of subreddits have gone private lately. Any chance you could run the bot again if you have time? Just to add the quality to the subreddits that has gone private, nothing else Trade (talk) 22:56, 14 June 2023 (UTC)

@BrokenSegue:--Trade (talk) 15:28, 15 June 2023 (UTC)

I think I would want to wait for the flux to end before updating the information. I think it's counter productive to have update a subreddit to "private" if it's going to become non-private again in 2 days. BrokenSegue (talk) 17:11, 15 June 2023 (UTC)

The "flux" ended four days ago --Trade (talk) 17:44, 17 June 2023 (UTC)

There is a complete list here. If you contact the creator it might be possible to find a way to make the bot "sync" with the website. That way Borked could simply remove the quality if any of the subreddits goes public again --Trade (talk) 19:46, 17 June 2023 (UTC)

@BrokenSegue:--Trade (talk) 19:21, 23 June 2023 (UTC)

Sorry for the delay. I really don't have a lot of free time right now and I'm not keen on importing data that will quickly go stale if we don't have any way to version this data. This entire bot task is kinda on hold due to the discussion at User_talk:BorkedBot#Social_media_followers_on_concepts which I haven't had time to resolve. BrokenSegue (talk) 19:20, 26 June 2023 (UTC)

No updates since October 11

Has YouTube discontinued its API? BorkedBot (talk • contribs • logs) is no longer updating.-- U. M. Owen (talk) 14:26, 24 October 2023 (UTC)

@U. M. Owen: no it must've just failed. i'll check it out. BrokenSegue (talk) 16:33, 25 October 2023 (UTC)

Ok I fixed it. The token it was using had expired. Should be good now. BrokenSegue (talk) 21:35, 28 October 2023 (UTC)

Perfect. Thanks for your work!--U. M. Owen (talk) 21:36, 28 October 2023 (UTC)

Miraheze article ID (P11250)

@BrokenSegue:

it would be nice if this bot could do what it does to Fandom article ID (P6262) statements, but also with Miraheze article ID (P11250) 😬 –Shisma (talk) 11:30, 5 November 2023 (UTC)

and also ‎Gaming Wiki Network article ID (P12143) –Shisma (talk) 12:41, 16 November 2023 (UTC)

Twitter: No populating of Twitter numeric user ID (P6552) on Twitter username since 5 June 2023

According to the contributions history of Wikidata, the last item BorkedBot (talk • contribs • logs) populated X user numeric ID (P6552), start time (P580), and point in time (P585) on Twitter username was Martin Ajdari (Q16919816) at 09:01, 5 June 2023. It would be great if you could please check this. Luamssuk (talk) 17:30, 22 November 2023 (UTC)

@ Luamssuk: this is intentional. Elon Musk shut down the API. BrokenSegue (talk) 17:36, 22 November 2023 (UTC)