Feature request: DQA toggle for multiple species in same obs

[Jane Widness]

This has been asked before, but I was thinking about it again recently and hoped maybe we could talk about it.  Can we please get a DQA toggle to flip if an observation has multiple photos of different species?  I often resort to using location isn't accurate when one photo shows an indoor houseplant and another shows a duck in a pond or whatever, but a separate toggle would really be better.  I also see lots of users, even ones who should know better, just adding the ID for the first photo in the set.  A DQA toggle might help remind people that you should only ID to the finest level that would include all the species in the observation.

[robert archer]

I would very much like to be able to flag the photo of the second organism so that it is not visible for users or importantly not shown in the photo browser. The Observer does not respond or is not active anymore. This means an otherwise good observation is ruined. 

[jdmore]

+1 from me -- I think that would be a very appropriate data quality attribute.

We would just need to be clear that it should NOT be applied in cases where multiple taxa are identifiable in a SINGLE photograph.  This could confuse folks.  (In fact, the attribute should be disabled and disregarded when there is only a single photo in the observation.)

In the latter case, the taxon being identified should be clarified in a comment (if not already in the poster's description or comment), along with the identities of the other taxa visible, if known.  And if appropriate, the poster should be encouraged to duplicate the observation for the other taxa in the photo.

As for flagging to "hide" other non-relevant photos attached to an observation (Robert's suggestion), my feelings are a little more mixed.  My initial action would be to add an ID to the observation that encompasses all the attached photos.  For cases where the poster is non-responsive, maybe a flag that curators could investigate further, and then possibly "hide" non-relevant photos in accordance with strict curatorial guidelines (to be formulated first).  I don't think I would want any and all users to have this power.

--Jim Morefield (jdmore)

[Lincoln Durey]

My (i hope/thiunk slightly better?) approach to these is to add an ID for the "Least Common Denominator" of all the organisms present,

so Beach trees and Oak acorns get "Fagales", Hickory and Taxoduim get "Tracheophyta", and plants and animals get "Life"

In this way the observation is still "Needs ID", the photos are sorted out of the more specific taxa and up to the higher level, and I leave a note:

You have several different species that are the focus of these photos.  Please split this into
one species per observation so that we can make an ID. (multiple views of a species are fine)

If possible I will tell them the ones I know.

I specifically use this quite often on observations I find that are already at Research Grade, but that have RG for the first image, and that species is nowhere evident in the other images.  Those superfuous images must get removed from the AI training.

Also, "Location inaccurate" may not be the best play, as they could be photos takes within a very small space, or might be days and states apart.  but my "Dicots" or "Plantae" IDs are always "correct" scientifically.

[Chris Cheatle]

While I appreciate the idea behind this, I fear a flag to force another user's photo to be invisible would be abused as a tool to 'remove' photos that are judged to be poor quality etc more often than it would be used for this.

[Mark Tutty]

I like the idea of IDing at the common level... and I hate the idea of hiding someone elses photos. The observations belong to them, and they are potentially putting them up with a purpose in mind. If we have too much power to impact on how people use the site, then we run the risk of alienating the very active users we want to encourage.

 

I can see merit in flagging a photo for exclusion from use in AI or the Photo Browser, but even then we could be excluding a particularly useful view because someone decided it wasn’t good enough. Very subjective, and we would all have different views on that.

 

We need to remember that the observations are theirs, and they are being kind enough to share with us their view of the world and what’s around them.

 

What I think would be cool, is if we could put “private fields” on other peoples observations. I’ll start another topic for this, because I think it is getting off topic...

 

cheers
Mark Tutty
kiwif...@gmail.com

--
You received this message because you are subscribed to the Google Groups "iNaturalist" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inaturalist...@googlegroups.com.
To post to this group, send email to inatu...@googlegroups.com.
Visit this group at https://groups.google.com/group/inaturalist.
For more options, visit https://groups.google.com/d/optout.

 

[Jane Widness]

To be clear, I never suggested hiding or removing anyone's photo.  The DQA toggle would do the same thing any other DQA toggle does: move the observation to casual grade until the issue is fixed.

[Mark Tutty]

Yes, sorry Jane... I did not mean to imply it was you! Keep in mind that this is a discussion forum, too, and no-one is going to judge anyone for putting up ideas no matter how crazy the idea might be, so even if it was you that suggested it, it wouldn’t be a mark against you 😊... I’m a huge fan of creative brainstorming, and sometimes ridiculous ideas trigger other ideas that lead to great solutions...

 

cheers
Mark Tutty
kiwif...@gmail.com

[Chris Cheatle]

Likewise, if it was not clear I was responding to the 2nd commentb in the thread, my apologies.

[robert archer]

The important issue here is that an observer may be absent for years or even dead. Five years or ten years from now there may be more of those. The ability to exclude these photos from showing in the photo browser that have been clearly added by mistake by the observer. The example I am thinking about is a plant or two, mammal and insect. If a careful curator cannot fix these by flagging - not removing, it will just means a potential valuable observation lost.

[Mark Tutty]

And Robert, I was by no means suggesting that it was a ridiculous or crazy idea!

 

cheers
Mark Tutty
kiwif...@gmail.com

[Tony Iwane]

I'm not sure adding a DQA would be a better solution than the current suggested protocol of adding an ID of a lowest common taxon (eg life, plants, animals). That should solve the photo browser issue and will add one more notification for the observer to see. I don't think photos which are within in the Community Guidelines should be hidden.

Tony Iwane

[Jane Widness]

The problem I've had with just adding a coarse ID is that often there are so many agreeing to ID the first photo that it goes to RG anyway.  So then I tick "Can be improved" and it goes back to Needs ID, and then people just add more species level IDs to it even though it doesn't actually need more identifications.  

[Mark Tutty]

Encourage others to also add higher level Ids. Have a snipet that you can copy and paste, I was going to give an example, but it’s better that you come up with your own wording.

[tony rebelo]

I think we should distinguish between observations with lots of different species, or only a few photographs of different species.  For those the common denominator would work best, but I would argue why make any identification at all?  Just make it "Life" and a comment as to the correct concept of an observation,  and leave it until it is fixed.  The DQA "mixed photographs" would take care of any 'research grade' issues.

What concerns me are good observations, where an accidental picture has sneaked in which is clearly not what the observer intended, or alternatively where there is a series of one species with an odd photo of another species which the observer misidentified while posting.  To ID this as highest common denominator is wasting good data.  iNature is not just about AI training, but also has lots of other spinoffs.   In these cases a DQA of "mixed photographs" would be more appropriate.  That keeps the intended ID available for those interested - after all it is a good observation of that species, while flagging it for those whose only interest is the pictures, that there is a problem with the pictures.

However, I would argue for a case of being able to flag just the offending photograph - a case of "this photograph clearly does not contain the same species as the community ID"  

I also concur with Robert.  People who post lots of pictures of different species on the same observation, and do not correct it within a week (or two) of being told, are likely to be one-offers, or dont-cares, or no longer interested or dead.    They are equally likely to be ignorant of locality issues, or identification protocols or that there is a dashboard that they can follow updates on.  These are data quality issues, not identification issues, and should be dealt with the the DQA section and not identification protocols.

It would be nice if curators could "explode" such observations to conform (and likewise if they could "combine" lots of single photograph submissions of the same organism at the same place at the same time; or duplicate observations for example of a pollinator posted but visiting some unusual/rare/conservation-concern  plant that the observer has no interest in), but there are all sorts of data quality issues that would need to be addressed for this to be an option.  Nevertheless I would still strongly support such a tool (with adequate guidelines) < but I would strongly oppose curators moving pictures between observations: that is a no-no! >

[jdmore]

I think we need to stop short of giving curators the power to "rearrange" the observations of other users.  What could possibly go wrong, after all?

The act of a curator flagging (or moving) a misplaced photo is really no different than that same curator expressing an ID opinion for that photograph.  And we already have a mechanism for expressing ID opinions.  So I have to agree with Tony Iwane that adding a higher level ID (and a comment, maybe tagging others to weight in too) is the best -- albeit imperfect -- solution we have without going down a very slippery slope of curator powers.

And maybe a well thought out DQA, such as Tony Rebelo's, Janes, and others' suggestions, could also be helpful for situations of abandoned accounts -- and maybe that is a decision for a curator to make in response to a user flag.