Structured Data Across Wikimedia/Image Suggestions/Feedback Commons

This page collects the feedback from the August 2021 mass-message initiative for the image recommendation features for experienced users, which is a tool currently in development by the Structured Data Across Wikimedia team.

You can read the full summary here.

From your experience, what is the hardest part about adding images to articles?

From my experience, the hardest part about adding images to articles is the arrangement of images in a way to make articles easier to read. --Soumya-8974 (talk) 11:41, 2 August 2021 (UTC)[reply]
Illustrating a wikipedia article by means of wikidata templates has the advantage of automatism, but concomitant disadvantage concerning the choice of illustration. --Havang(nl) (talk) 11:58, 2 August 2021 (UTC)[reply]
I've never found it at all difficult, but the main thing I think would make it easier for others would be to be more explicitly presented with the relevant choices: e.g. thumb vs. gallery vs. including the image in the template. The fact that some methods require explicit "File:". some have it optional, and some don't allow it is ridiculous. It would be much simpler for naive users to be able to fill out some sort of form to build the appropriate wikitext. However, I think it is very important that it remain wikitext, or at least be able to be serialized into wikitext. It has been a major problem on Commons that we can't do diffs across edits that change structured data. - Jmabel (talk) 15:40, 2 August 2021 (UTC)[reply]
У меня проблемы возникают при обновлении изображений в статьях, когда изображение "закопано" в Викиданных. Обновление изображения в Викиданных - уже проблема. И после обновления часто появляется старое изображение вместо нового. Такое впечатление, что Викиданные не дописаны, бета-версия. В принципе как программист я против Викиданных, это лишняя надстройка. --Matsievsky (talk) 19:03, 2 August 2021 (UTC)[reply]
The biggest issue that I've run into is in writing captions in other-language Wikipedias. The interaction between left-to-right languages such as English and right-to-left languages such as Arabic sometimes makes it very difficult to write captions, because when I paste in translations, the position of the cursor changes. Sometimes text is placed where I don't want it; or it might be split up around text already present in the caption.Jacqke (talk) 11:28, 3 August 2021 (UTC)[reply]
There are three issues for me. First, changing a misplaced image in a Wikipedia where I do not understand the language (except by automated translation) is difficult, because the context should remain the same so that no errors are introduced. Removing an image (which may be necessary) will possibly sound the vandalism alarm. Second, some Wikipedias use localized syntax for images and taxoboxes that are sometimes missing or wrong. If I cannot find any good examples, then I might be unable to add any images here. Third, I do not presently know how to deal with errors introduced from Wikidata, typically when several images are given for one object (e.g. a biological taxon), but just a single caption, which does not match the top image. See e.g. Commons:Category:Brassicaceae, where an image of a flower is displayed on the right side which is certainly not Barbarea vulgaris (caption). --Robert Flogaus-Faust (talk) 14:16, 8 August 2021 (UTC)[reply]
Its probably easier to add images into infoboxes since no additional formatting or prefix is generally needed. Crouch, Swale (talk) 10:46, 4 August 2021 (UTC)[reply]
For me it was always easy to add images to articles on Wikipedia - Richardkiwi (talk) 15:55, 4 August 2021 (UTC)[reply]
Personalmente non ricordo di essermi mai trovato in difficoltà nell'inserire una o più immagini in una voce su wikipedia, semmai alle volte è un po' frustrante doverne scegliere un numero limitato per esigenze di impaginazione, cosa che mi trova assolutamente d'accordo ma che rischia da lasciare nell'oblio tante altre immagini che potrebbero essere utili per la comprensione e approfondimento del testo. Soluzioni tecniche, come l'oculata aggiunta di una galleria di immagini o la possibilità di scorrere più immagini nello stesso riquadro, sono alle volte sovra(e mal)sfruttate nel primo caso, da chi è desideroso di mostrare il '''proprio''' lavoro dimenticando che il fine è mostrare la miglior immagine didascalica possibile, la seconda perché non compreso nel funzionamento dai lettori (molto) occasionali.--Threecharlie (talk) 17:07, 4 August 2021 (UTC)[reply]
The most difficult thing for me would be to place the image where I want it, i.e close to the relevant text. It's still a mystery to me why the images sometimes end up where they end up. --Judithcomm (talk) 00:06, 11 August 2021 (UTC)[reply]
I'm happy with the system for adding pics to articles. I do note that it is not uniform and takes a while to learn. One problem is that not all the images are in commons. Some are still in en WP or in articles of other languages. Second, there is still some inflexibility of machine formatting. You may do a good design at one width only to find that a slight change in width throws everything out of place. For the galleries, well, we need them. It is the only way to cluster a bunch of pics you want at a certain location. Pictures should have a meaningful location but I do not know how to get that across. Many well-meaning editors go through and disorder all the pics because they look more aesthetic somewhere else though meaningless there. How do you deal with that problem? Searching all the pics is something of a problem. It often does not find them even though they are there.Botteville (talk) 14:35, 29 November 2021 (UTC)[reply]

Incorporating machine-generated image suggestions into the editor workflow: concerns and potential

We have provided a few options for how we might incorporate machine-generated image suggestions into the editor workflow.
As a Commons user, you might be particularly interested in option #5, where we are exploring notifying users who have uploaded an image if the image(s) they’ve uploaded are a match to any articles on wikis in the language(s) the user speaks. What concerns do you have about these options? What excites you about these options?

General comments

My main concern is that the notifications will be annoying, especially for people with a large number of image uploads. Commons is an independent project from Wikipedia. People who upload images to Commons are not necessarily interested in contributing to Wikipedia, or may be uploading images specifically for work on another project like Wikisource. Any such notifications to image uploaders should be opt-in so that only those uploaders who are interested in adding images to Wikipedia will be shown them. Nosferattus (talk) 12:34, 2 August 2021 (UTC)[reply]
All I can say is that these recommendations make me even more glad that I never use the Upload Wizard. - Jmabel (talk) 15:48, 2 August 2021 (UTC)[reply]
If I manage to get the owner of an image to submit it to Commons, the Wizard is functional at the moment. But if it were to start asking them to make additional choices, it might discourage them from uploading. Please keep it simple. Perhaps a link or tab to reach the recommendations after the image is uploaded? Jacqke (talk) 12:11, 3 August 2021 (UTC)[reply]
Option #5 might be really annoying. However, if suggestions pop up only when there are matching pages without images, then this feature could be helpful. In this case, an option to show all matching pages without images in any wiki irrespective of language might be even more helpful for me. However, his option should be opt-in only. --Robert Flogaus-Faust (talk) 21:00, 3 August 2021 (UTC)[reply]
Mi allineo a quanto affermato dai colleghi qui sopra; per un utente esperto potrebbe essere più una scocciatura che un'utile opzione, per uno non esperto un ulteriore spaventoso ostacolo che lo farebbe desistere da ulteriori upload. Già ora si deve combattere con l'ABC, e l'oltre un milione di immagine non categorizzate tra quelle caricate nel 2021 dovrebbe far riflettere perché l'utente non perde nemmeno tempo a mettere una categorizzazione utile minima (o un nomefile e/o descrizione sufficientemente precisa da essere trovata, sai quante chiesa infilate nella macrocategoria Churches ci sono? da piangere...)--Threecharlie (talk) 17:07, 4 August 2021 (UTC)[reply]
I frequently use the Flickr and Geograph upload tools and I'am already happy if they both work as expected. (Geograph doesn't at the moment.) One type of description for one upload is more than enough for me. Make it a mandatory field if you like, like the date, but don't mess with it any further. One possible improvement would be to have a pop-up thesaurus (cats tree) to choose from for the categories, based on the subject, but please keep it optional. --Judithcomm (talk) 00:19, 11 August 2021 (UTC)[reply]
The past week I was pleasently surprised to find some brand new photo's on Commons of a subject I've been writing about on Wikipedia. I found them by accident and immediately added one to an article. Would it be feasible to create a 'google alert' service to Commons that warns you about possibly interesting uploads? Based on the file name, the categories, but maybe also on the discription? With a completely free format list of search terms? With options for how often you want to be alerted? I would love that. --Judithcomm (talk) 00:55, 11 August 2021 (UTC)[reply]

Upload Wizard Improvements

What concerns or excites you about improved copy, alerts and messaging in Upload Wizard to encourage more specific depict tags?

...

Notifications

What excites or concerns you about notifying users about potential image suggestions for an article (see image 4), with the ability to opt in or out of those notifications?

...

What excites or concerns you about notifying users who have uploaded an image to Commons (see image 5) that the image(s) they've uploaded are a match to any articles on wikis in the language(s) the user speaks?

...

What excites or concerns you about a tool that allows users to review suggestions and add images to the articles (see image 6)? What might be missing in this tool?

...

Article Talk Pages

What excites or concerns you about placing automated messages on talk pages of articles that have a potential image suggestion?

Please don't. Talk is for people, not machines. If I have to wade through lots of automated messages to find the ones from other users, I will stop visiting the talk pages. --Judithcomm (talk) 00:23, 11 August 2021 (UTC)[reply]

Visual Editor

What excites or concerns you about adding suggested images to the search UI in Visual Editor (see image 9)?

...

What excites or concerns you about placing a blue dot notification by the "insert" dropdown in Visual Editor when there is an image suggestion available (see image 10)?

...

How can we help users add appropriate captions?

Is this really a problem? I have rarely, if ever, seen an image with an inappropriate caption, except in an article that was almost entirely incompetently written. I will add, though, that if the system is going to suggest adding an image, then it probably needs to suggest an appropriate caption. However, keep in mind, Commons has been plagued with structured data that misses the point, and would lead to misleading captions: e.g. a photo of a city street that has a "depicts" tag for a stop sign, which is barely even visible in the image, or a "depicts" for a place, when the picture happens to have been taken there but in no meaningful way depicts the place, or for a person who is very incidentally in the photo. - Jmabel (talk) 15:46, 2 August 2021 (UTC)[reply]
I would suggest a text-box that asks users to write sentence or two, addressing who, what, when, where, why and how for the image and also for its subject, as best they can. The box should also include a sentence about how addressing these points will avoid confusion and make their image useful. Jacqke (talk) 11:53, 3 August 2021 (UTC)[reply]
- I agree, especially on the What, When and Where; Who needs perhaps an explanation (who is depicted / who is the artist in other cases) and Why and How only in very specific cases. So with the first four I would already be very happy. JopkeB (talk) 12:37, 4 August 2021 (UTC)[reply]
Non saprei bene come ma pregandoli di essere specifici (vedi il mio commento sopra) ci eviterebbero tanto lavoro sporco. Benché non possa smentire chi mi ha preceduto nel ritenere che chi carica abbia poi effettivamente intenzione di sfruttarle in qualche progetto wikipedia, una sorta di doppione di un Flickr, spererei che chi arriva in Commons lo faccia almeno per la convinzione di rilasciare contenuti multimediali con licenze molto libere, ma spiacevoli episodi in cui chi rilasciava le stesse da siti specialistici si è pentito (anche perché si pensa sempre prima all'io che non al noi) mi fanno rimanere, seppu moderatamente, disilluso.--Threecharlie (talk) 17:07, 4 August 2021 (UTC)[reply]

Re: request for feedback on Commons

The largest problem I find with newly uploaded images is that far too many contributors do not say where the image was taken. This is vitally important information for any scientific analysis of images, particularly in biology-related topics. Adding location information needs to be made a more integral part of the upload form, and with some form of prominent reminder before completing the upload (something like "you have not completed the location information, are you sure you don't wish to?"). - MPF (talk) 10:09, 2 August 2021 (UTC)[reply]

I agree. It has to be recommended to have the location as part of the filename, and/or it should be good to have a question abiout the location during the upload process, similar to the date question. As times of creation and of upload are already generated during the ulpoad process, I name files by location and object (or sometimes for scanned images source and object). -Havang(nl) (talk) 11:53, 2 August 2021 (UTC)[reply]

I totally agree that, one way or another, the location should be added to each file. Filename is fine, or part of the description (indeed like date), geocoordinates or a category with the location. Not only for biology-related topics, but for all sort of topics. The category structure is for an imported part based on countries and places, so also therefor we need locations. JopkeB (talk) 15:03, 2 August 2021 (UTC)[reply]

I agree that appropriate location information is important, but this is frequently just a part of the description and / or some categories. If the upload wizard can actually recognize location data other than geocodes, then the suggested warning might be helpful. Otherwise the warning would be just annoying and rather pointless. Anyway, identification of organisms is frequently difficult if the organisms are not properly identified by the photographer. For example, if you take a photograph of an inflorescence of a plant from the Apiaceae family so that you can see only the flowers, identification may be impossible. If you take a photograph of a plant cultivated in a botanical garden without providing the correct ID, then this plant will almost certainly never be properly identified even if the precise geolocation is provided. --Robert Flogaus-Faust (talk) 21:44, 4 August 2021 (UTC)[reply]

I think the image metadata on Commons would be better quality if adding metadata was integrated into the upload process, rather than awkwardly tacked on at the end. For example, the date and license could be automatically generated from the existing form inputs. Nosferattus (talk) 12:49, 2 August 2021 (UTC)[reply]

There are some bots that do things like that on Commons, e.g. Commons:User:BotMultichillT. AFAIK this works nicely. --Robert Flogaus-Faust (talk) 21:14, 3 August 2021 (UTC)[reply]

As far as I know, a bot can only convert data into categories or to Wikidata if the data already exist, not if they are not in the file at all. I think that is the problem. JopkeB (talk) 03:52, 4 August 2021 (UTC)[reply]

I recently (a few weeks ago) saw an IP-user upload a file with a common name, I don't know exactly, but something like garden. Then they requested a rename for (p.a.) back garden. Declined it and asked the user for more information of the photo that was taken that day. He didn't understand. My follow Page is big, but maybe I didn't follow it (I forgot, I guess). It would be nice if a bot would pick out things like that and put them in a category 'to check', although I think it will become a big category. If that will solve a problem, I don't know :-) - Richardkiwi (talk) 15:55, 4 August 2021 (UTC)[reply]

Come ho già scritto, riprendendo chi mi ha preceduto oltre al problema della non categorizzazione, anche la presenza di descrizioni banali crea problemi irrisolvibili, e il lavoro sporco di chi si illude che categorizzare un'immagine di una chiesa (che è evidentemente identificabile sia una chiesa) se resta File:Chiesa o anche File:Chiesa parrocchiale, File:Parrocchia di San Rocco e alla via così infilarle in una Category:Churches non serve a nulla. Attenzione però, anche io recentemente ho caricato almeno una foto di una chiesa che, non avendo evidentemente abilitato il GPS, non ho la più pallida idea di dove sia e ho perso ore setacciando G.Maps nella speranza di ritrovarla... quindi non tutti sono demoni O:-) --Threecharlie (talk) 17:07, 4 August 2021 (UTC)[reply]

I've seen many uploads with silly, meaningless filenames, like 'amamamamama', 'first film 01' and the like. Would it be possible to detect these during the upload process and suggest to the uploader that they give the image a more appropriate name? I'm always hesitant to rename these pictures later myself, because sometimes they are part of a series and the fact that they share one goofy name tels everyone that they were probably taken at about the same time and in about the same place, which can be usefull if no other info is available. --Judithcomm (talk) 00:44, 11 August 2021 (UTC)[reply]