A place for discussion about anything related to the Recogito annotation platform
August 2, 2016 at 2:07 pm #1716
This is a query, leading to a possible recommendation, if there is not yet an easy answer to it.
Some of the maps in Recogito (in particular Arabic maps such as Al-Qazwini’s Cosmografia) have had rectangles drawn around many or all of the words on the map—in some cases wrongly, I might add—but no transcription given or georesolution carried out. Leaving aside the question of whether there is any point to such “annotations,” it does make it harder for another user to come along and transcribe or resolve these labels. The semi-transparent red square often makes it hard to read the text underneath, especially if the text is already obscure, handwritten Arabic and a low -resolution image.
Would it be possible to offer an option to hide the annotations altogether, either one by one or for a whole map, to make it easier to read and work on? (This would also be useful for when the box around one annotation partially obscures a second word.) Otherwise, sometimes the only way to read and annotate a word is to delete the existing annotation, read it, and then draw the box anew.
Thanks for any advice on this.
August 2, 2016 at 2:17 pm #1717
August 2, 2016 at 2:22 pm #1718
yes, an option to toggle the annotation layer for the entire page is already scheduled for Recogito v2.
As for the rest of the new image annotation UI: that’s not designed yet. But as we’re progressing nicely right now, I’m thinking about starting work on this right after I’m back from vacation. (Still here this week, and then away for two weeks.) Hiding other/overlapping annotations while working on a specific one definitely makes sense, so I’ll see that the UI allows for that. (And feel free to post additional thoughts, ideas, input etc. about other useful image workflow & feature enhancements you can think of here or on GitHub!)
August 2, 2016 at 3:41 pm #1721
Thanks, @Damien and @Rainer. That’s good to know.
Am I right about there not being much point in drawing a box around a word and then *neither* transcribing nor georesolving it? (Especially if the box is around each word, which sometimes isn’t even a complete placename!) Should there be guidance about that, or even (radical) imposed by the software? If you draw a box around something, but add *neither* a transcription *nor* a resolution, then you can’t save it?
August 2, 2016 at 3:54 pm #1722
Not that I’m accusing anyone… …but I had wondered whether the contributors’ leader board was being ‘gamed’. We did discuss the advantage gained by making lots of ‘inadvertent’ errors, and having to correct them. Creating lots of empty annotation boxes is an even quicker way to rise to the top.
I feel embarrassed even to speculate about such things among this high-minded community.
August 3, 2016 at 8:40 am #1723
@gabrielbodard: hm – not sure about that. True, it’s not an incredibly valuable addition. But we always envisioned some sort of additional “crowdsourcing frontend” that would crop (and rotate) the “boxed” image segment, and then e.g. post it to twitter, asking “Can you help with this toponym”. @leifuss has been working on some code for this as well.
@damien-bove: What’s this – the American elections? (“Russia, I hope you can find Hillary’s missing Recogito edits”) Jokes aside: yes, the old Recogito was pretty casual in terms of what counts as a “contribution”. (Everything did – including deletes, and text selections made automatically by re-applying a user-made selection across the whole text.) Recogito 2 now records contributions with more structure. So it will be entirely possible (and up to us) to define what metrics to use for the scoreboard in the future. (I still think a scoreboard is essential 😉 )
You must be logged in to reply to this topic.