User:Secretlondon/Wikilovesmonuments

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search

Wikilovesmonuments is an annual competition to upload images of listed buildings. However it seems to be disorganised and is 100% opaque. I've no idea how the judging works but the organisation of images isn't organised. At the time of writing this there are over 10,000 images from previous years (UK only) which have not been catalogued properly.

Discoveries[edit]

The project is presented as a map called monumental (https://maps.wikilovesmonuments.org). Each listed building is plotted as a pin. The pin is red if we do not have a photo and is green if we do.

The pins come from Wikidata items. Each listed building has been imported as a Wikidata item, using Historic England's naming system. It's a data dump, and if we already have a Wikidata item for the building (say if it's a well known building) they may not have been merged so we have two entries not one.

Monumental looks for Wikidata items that have the image statement. If there's an image the pin is green, if not it is red. If we have photos on commons but nothing attached to the Wikidata item the pin is red. If we have a whole commons category of images of the thing, but no image statement, the pin is red, even if the commons category is linked to the Wikidata item.

Key Wikidata statements are image, and commons category. I presume the map is drawn from items with heritage designation and identifiers like National Heritage List for England number (see https://www.wikidata.org/wiki/Q26525707 as an example from this year). I added the image statement so that pin is now green.

Issues with item selection[edit]

If the listed building does not exist it is still plotted. I haven't worked out how monumental selects items but former listed buildings are being picked up. Shanklin Pier (https://www.wikidata.org/wiki/Q105037739) was demolished many years ago but is still on the map, as it's set on Wikidata as a listed building which ended. I personally walked to the site of a stink pipe (or something) this year to find it was no longer in existence.

Scottish buildings seem to be pulled from several sources. If they disagree, or say that it's not actually the earthworks for an ancient monument after all, it's still been imported and plotted as though it is.

Issues with software[edit]

Monumental fails silently. If you upload your images via the map from your phone it seems to lose over half of them. There is a bug about this but there's been no action on it for about two years (https://phabricator.wikimedia.org/T175676). I lost lots of work this year. I think it is due to the size of files but my files were nothing special.

Monumental knows the Wikidata item for each site but photos uploaded through it are not linked to it, so each one needs linking manually. This is/should be a bot job.

Secretlondon's organisation process[edit]

There's no public topdown organisation but I think the priority is that we make the pins green if we have a photo. This stops people having to track them down. The images also need to be linked into commons local categorisation schemes so they are not separate from other local photos. Commons images are organised by categories and by structured data and I try and do both, but my categories are basic due to time, and lack of local knowledge. Subcategorising is another job.

I am working from Category:Potential_image_for_Wikidata_item and the petscan links.

1. Load each image in turn

2. Add to a local category. The Historic England naming is by council area (for example Torquay photos are uploaded as Torbay). I do this simply to stop the images being disconnected from other local photos uploaded outside of wikilovesmonuments. It's another job to subcategorise them properly.

3. Go to structured data and link to the Wikidata item. It should be depicts whateveritis. This doesn't work from commons a lot of the time as it only shows you 7 possibles. If your photo is of something with a common name it probably won't find it. In my experience the most reliable way of finding the Wikidata item has been to search on the listing number using the Wikidata website. You then copy the Q number into the depicts box on commons.

4. Once you've added a depicts statement you need to refresh the page or anything else you do to the commons entry will fail.

5. Sense check the thing. Some images are not of the thing they are supposed to be. I don't have time to do a deep dive on everything but somethings are obviously wrong and the linked entry on English Heritage's website provides a description. If its clearly wrong I've removed any depicts, the listed building template and made a comment on the description. All the entries for a particular bit of Canterbury Cathedral this year were just general ones of the cathedral.

5. Go the Wikidata entry and add an image statement with the name of the image file (without the file prefix). You can do more if you want - check description, add other names, check it's an instance of the right thing.

6. Go back to the commons photo and take it out of the category Potential_image_for_Wikidata_item.

7. It's done! Pin is green. Image off the list!

Comment[edit]

My real issue is how opaque and unwiki it is. If you want to get involved you have to work out how it works. There's no public documentation. There was a comment page on a website (https://www.wikilovesmonuments.org.uk/faq) but comments were moderated and were not unmoderated or answered for most of the month.

I've tried to look into this more. Monumental feels unsupported. If you try and report an error it sends you to either github or Phabricator, depending on whether you click on the main map, or an individual entry. Bugs don't seem to be managed in either case.

I can see where Wikimedia UK have reported issues on monumental and there's been no action. Apparently it's plotting SSIs as listed buildings. It might be using the presence of Heritage Designator as a statement on the Wikidata item.