Enhancing annotation: Adding tags to selections

Hi Tropy community!
Wouldn’t it be useful (to other people than myself) to be able to add tags to image selections and not only to the image itself? Tagging people in a text, highlighting a paragraph where a certain concept appear, tagging unreadable words to collect them, etc.
Interesting corollary: in the global view, when a tag would be activated, the search results would be directly centered on the image portions (all the signatures, all the unreadable words, all the bibliographic references, etc.).
Tropy would then become a free and fresh alternative to MAXQDA and other annotation software.

2 Likes

Good question!

At the moment, tags can only applied at the item level (not the photo level); we have considered making it possible to tag any or all of photos, photo selections, and notes independently but decided against it mainly to reduce UI complexity. It would certainly be possible to add tags at lower levels (an item’s tags would then have to be defined as the sum of its own as well as all the tags nested below it), so it will be helpful to hear your thoughts.

Many thanks for your answer, and good to know that you did already discussed this question.

[I didn’t tried the photo grouping functionality yet, so I hadn’t noticed that the tag was only available at item level]

I understand your concern about the UI complexity. In my opinion, and having rather complex archives (but I presume every researcher will have its own complex questions/material), a software that makes it possible to tag only pictures as a whole is not much more useful that simply using - on a Mac OS - the Finder tags. It’s already pretty common to use folders to sort a picture collection and add colourful tags to distinguish the items: Tropy is not really adding anything (the “annotate your photos” slogan seems to me a bit overrated).

I can not imagine all the uses, but think for example of someone who has a collection of old photographs: tagging people makes sense if it is possible to frame / select their faces. For now, and this is the example of your video, the annotation is used to transcribe a portion of a document, but one could imagine having several “categories” of annotations: transcription, identification of concepts or people, personal notes or comments, identification of a type of media/element (an illustration on a newspaper page, a signature, an address on a letter), etc. We could also automatically give the selection box the color of the tag assigned to it, to make reading faster, or display the color tag in a corner.

Of course, I do not know how much you would like to develop Tropy in the direction of an analysis software (and therefore no longer only a collection software), but one could very well imagine researchers who would manually tag their corpus to then count occurrences or co-occurrences. It could be very useful on small to medium corpora on which the OCR is not possible.

In short, I do not know if all these ideas deserve an answer today, but they are things that would make Tropy’s use justifiable in my case.

1 Like

I’d really appreciate tags at the selection level too. I hope this is introduced at some point. I understand the need to keep the UI simple, but perhaps “tags for selections” can be enabled by users who need more complexity.

Enabling tags at the photo and selection levels would be relatively easy; the bigger UI problem is the more general ‘searching within an item’. So I’d be curious to know how you would intend to use selection-level tags; if you’re only concerned with being able tag selections to highlight them in the list or to export the information to some external data repository, this is probably something we could add quickly.

However, we don’t have a clear UI solution yet for searching within a given item or displaying partial hits within an item – so in terms of searching there currently would not be much difference between a tag applied to a selection versus one applied to the full item.

Here is my use case: I work with image plates of mixed fragments from several ancient scrolls. I also work with images of individual fragments. I only want to tag an item with a fragment tag if it contains just one fragment. Otherwise, I want to use fragment tags on selections. I would like to be able to tag a fragment everywhere it appears and be able to call up all the items and selections of a fragment in the search results. I want to review, compare, and take notes on each item and selection containing a particular fragment in the search results.

Also, I’ve been naming the selections according to the tag, but it’d save a lot of time if I could just apply the tag to the selection to name it.

Thanks for considering this.

Would also greatly appreciate this feature. Regarding the results UI, it would be fine to display item tags and selection tags together in a list, in a very similar way to how items are displayed in the results UI now. The only difference being tags on selections would immediately take you to the image and selection within the item when selected. No need for displaying a hierarchy or anything like that.

To clarify, to display a selection result from a tag, display the item, perhaps with some icon or text to show that it is a selection within that item. If, for example, there were two selections with the same tag within an item (“egypt”, for example), selecting the “egypt” tag in the left pane would display the item twice in the main pane, with some kind of indicator overlaid on the item icon to communicate that these are tags attached to a selection within the item.

This would be perfect. Searching within an item is not necessary.

The main problem is when you have an item that spans 10 pages, and there is only one sentence related to a tag, say on page 7. When you come back to the document, you have to scan through the document to find the one sentence related to the tag you’re interested in.

To add to this: When our notes on the 10 pages are few, we can create a note with just one or two words, and problem solved. But if we have a 500-word transcription per image, say, that’s when the scanning for a key concept or name becomes tedious.

Yes, this is my case exactly. Scanning a 10-page item with several hundred words per page for a single topic or name associated with a tag is time consuming. The ability to tag selections of photos within an item seems to me the only feature that hinders Tropy’s usefulness.