User talk:Fæ/Project list/Fleuron
Hugeness edit
@Pigsonthewing: later today there's going to be a pause in this upload for several days while I'm travelling. I'm pondering whether it's worth uploading all the files that we can if their value may be debatable. The error rate is probably more than 5%, though not all of the non-Fleurons will be deleted as they may have their own value. We can see many 'duplicates' of the same printer's Fleuron being scanned in different uses, this could be useful as some clever person could use many printed copies to restore a digitally enhanced 'ideal' version, but that presumes we can collate close duplicates.
Currently there are 39,000 images uploaded and the run is at "A n" in alphabetic order by book title. That's "A n">"Ab" so just in the "A" collection we are only 58% done based on book title count (not image count). "A" may be the largest letter, but I would hazard a guess that "T" is going to be equally huge. I'd appreciate it if you could take a browse through what we currently have, and see if you have any observations about how to rationalize the upload more than we already do. If we stay as we are, we may end up with a couple of hundred thousand images, with no easy way to navigate them for reuse. Thanks --Fæ (talk) 11:28, 17 May 2017 (UTC)
- @Fæ: Thank you for your work to date, on this. I take a fairly liberal view regarding the addition of such images, but then, I'm not doing the work ;-) I wonder whether the discoverability and usefulness of the images could be enhanced by applying heuristics to add further categories - one per book, where there are more than a handful of images for instance. Perhaps also by format (landscape-format images of fleurons; square (ish!) images of... etc.) Some fleurons, like File:A companion for a sick bed- or, a preparation for death Fleuron T119682-4.png, include images of objects, like birds. I wonder whether we could partner with Cambridge to crowd-source the tagging of these? I'll conatct them about that, once I return from WikiCite. Then there's the issue of how we re-categorise non-fleuron images. Andy Mabbett (talk) 10:04, 18 May 2017 (UTC)
- I'll re-launch the uploads, but it would be helpful for you to drop in to my user talk page if anyone starts to make a fuss about this project there. As the fleuron+year categories were created for this project, I don't see there being an issue with complaints about 'flooding'.
- I take your point about the books, and as the book short code is in the filenames, a future housekeeping task could fairly easily add new book related categories. However my IA experience demonstrated that creating good book categories is a bit of a nightmare, so it would take a fair amount of testing and re-writing for useful categories to be made reliably. Were anyone to get excited enough about this, it would make for a good Wikidata project, i.e. import all the book titles, short-codes and other data to Wikidata, then tag all the images with the relevant Wikidata record based on the short code. That's not me though, it's the sort of project I'd rather get paid to sort out than suck up my limited best free volunteer time. --Fæ (talk) 12:30, 31 May 2017 (UTC)
Non-fleurons in the fleuron set edit
The fleurons you auto-uploaded look like they were extracted automatically from the books without thought to whether all non-text content would really be fleurons or not – see for example this image. In general a high percentage of these categories contain non-fleurons. It's not clear how to fix this except for by hand, but it does make these categories a lot less useful.
Ilzolende (talk) 18:21, 30 May 2019 (UTC)
- True. Cat-a-lot is our friend. You can quickly mass add poor quality duplicates to Category:Uploads by Fæ needing speedy deletion and obvious non-fleurons can be removed from the fleuron category without contest. In this way hundreds of images might be shuffled around in a short editing session rather than a handful. --Fæ (talk) 19:15, 30 May 2019 (UTC)