Merge existing duplicate observations *and* the identifications and comments on each

Description of need:
Like many have done I have several past observations that have been uploaded more than once. In some cases both uploads have their own chain of identifications and comments.

Questions about this have cropped up in the past, but I have never seen any feature request, and at present there is no way to do this without deleting one of the observations, which results in the loss of the associated identifications and comments on the observation deleted.

For those of us who would like to clean up/neaten past observations and to retain important data at the same time this is important.

Feature request details:
Ideally it would be a simple matter of A) merging existing observations and B) merging the identifications and comments as well.

If the images are the same, then one image is retained, and if one observation has more images than the other then the identical images are pared down to one image, and the others retained.

Comments and identifications would be merged, and if the same person had provided an ID for each, then only one ID suggestion for that person would be retained.

This would be, in some ways. very similar to how topics in the forum are combined/merged (or split)

This has been somewhat addressed in topics like the following two, but not to any satisfaction as the answers are either, “ignore it and do better next time,” or “you can’t do it, delete one and if the photo is re-upload it to the other observation.”

Good suggestion. Worth addressing is what will happen to the URLs of IDs, comments, etc. of the observation that isn’t kept. Will they just cease to exist, or are you looking for a mechanism where such links will now point to the observation that is kept?

2 Likes

Since part of the suggestion is that comments, and IDs be merged as well, I suppose that a redirect pointer would probably be how that is handled.

How is it handled when a comment in a forum thread is moved to or merged with a new topic?

This sounds like a good way to deal with the problem, but I’ve noticed that with some observations where multiple images are split up, one photo may be more easily identified than the others. What would happen if the same identifier had identified two different photos of the same organism differently?

My (uninformed) guess is that if the comment and ID timelines for the two observations were merged, whichever ID was more recent would “win”. Which might be a problem if an identifier took a look at the observation with the good photo, quickly added or confirmed an ID, and then took a while to come up with an ID for the observation with the more challenging photo.

So if this goes through, I think it would also be good for anyone following either of the observations to get a notification that observation A has been merged into observation B

1 Like

I agree that people who interacted with the observation should be alerted to changes. That seems like common sense.

In terms of the merging, I’d think that the person merging the observations would select which way the merge went (A → B or B → A). The primary photo would then be based on whichever observation adsorbed the other.

As with any other observation you’d be able to edit the final merged observation to rearrange photos, or to add new ones.

1 Like

Maybe I’m missing something, but if two observations have different photos, I generally don’t think of them as “duplicates.”

Although if they are different photos of the same individual organism taken at the same time, then yes, I agree that they should have been posted in a single observation, and that merging them with comments and IDs might result in unintended association of the comments and IDs with the newly merged set of photos.

So might be best to implement at the same time as this feature request:
https://forum.inaturalist.org/t/have-system-add-a-comment-when-photos-or-audio-change/14985

1 Like

Possibly they mean that one observation has additional photos, or has a different one placed as the primary photo.

One observation might have photo order A, B, and another duplicate observation might have photo order B, A, C.

If photos A and B are different quality, or show different details, then it would be possible for the same identifier to wind up providing different identifications for the two posts.

3 Likes

If it were my ID / comment - I would prefer a notification - so I could look at the ‘fresh’ obs with new eyes.

If the leaves and flowers were originally two separate obs, they may well have drawn very different IDs. All the pictures together … now it is obviously This species!

Diana says it is this species - may well be very wrong, if Diana has a chance to look at it as it now appears on iNat.

If it were my own obs, I would copypasta, @mention, then edit as appropriate.

Merging or separating out threads of comments in the forum is different. Merging obs is also going to affect what goes to GBIF?

1 Like

No more than adding/changing an ID does when it’s at a level where that makes a difference, or that removing an observation does.

In the long run it should make GBIF data derived from iNat more accurate.

In addition, if IDs and comments are merged, there will be instances where a combined observation now has the necessary IDs to become ‘research grade’ when it didn’t previously, making the observation available for GBIF. Of course there may be instances where the observation gets bumped back to ‘needs ID’ as well.

1 Like

Maybe I’m missing something, but if two observations have different photos, I generally don’t think of them as “duplicates.”

I’ve definitely come across series of observations where the observer either didn’t realize they could add multiple photos to the same observation or there was some sort of uploading error, because each observation is very clearly the same individual at the same time, just with a photo from a different angle.

4 Likes

It would be nice to have a feature to merge and fix observations based on several images of the same individual. This seems to come up every now and then (latest instance here). I consider these “split observations” rather than straight-up duplicates. I imagine one way this could be done is by having a function in batch edit mode where one could select multiple images and click a “merge selected” button, for example.

For organisms with several identifying features spread over multiple observations, having the pictures together in one place would help with IDs. As an identifier, I’ve experimented with an “identical observation set” field as a way to collate several observations that were taken at the same location and time and obviously the same organism. It requires attention to details and is a bit of a guessing game with unresponsive observers, and not every observer allows others to add observation fields.

1 Like

Hello, that’s me you’re talking about. The workflow I use to upload observations is very clunky and likely to remain such until iNat Next app comes out for real. It basically leaves me no way to combine photos into an observation before submitting them. I would love the ability to merge them on the website later.

some observations where multiple images are split up, one photo may be more easily identified than the others. What would happen if the same identifier had identified two different photos of the same organism differently?

So, to put it into perspective, think of what happens when an observation with a single photo features multiple subjects — such observations are duplicated and identified distinctly, right?
Same could be done with “gallery” observations and would basically still be in the spirit of the former, except the photo search would come up with some effectively misidentified images, containing no queried subject at all.
That’s not good.

So you raise a valid concern.
Along with merging functionality, we’ll need benefit from splitting functionality.

To clarify, I don’t think @m_whitson is talking about cases where different species were intentionally put into a single observation, but where two very similar species were erroneously thought to be the same one and later recognized as distinct. Sure, it’s supposed to focus on a single subject, but things don’t always work out that way.

It could be said that cases like that are rare enough where such observations could be manually split by duplicating and then removing images from each duplicate observation… So splitting functionality isn’t strictly necessary, I have to retract what I wrote before editing…

There is already a ‘duplicate observation’ function that can be used for the splitting action.

Duplicate the observation, make a comment specifying which individual (if more than one species are in a single photo), and delete any extra photos in the duplicated observation.

1 Like