Add a way to search for IDs which are both improving and explicit disagreements

image
Speficif requests : Is there a way to retrieve all the IDs for a defined taxon that “disagree” with other IDs ? Especially if they are assigned the “improving” category ?

For example, an URL request could be a leaderboard like this (based on suggestions by @pisum in other topics):

https://jumear.github.io/stirfry/iNatAPIv1_identifications_identifiers.html?taxon_id=47792&category=improving&disagree=true

where disagree=true would be the desired option not existing yet. It would be great if there were an equivalent option unknown to me.

Or another URL could be :

https://www.inaturalist.org/identifications?taxon_id=47792&disagree=true
inspired from the URL to identify “withdrawn” IDs that follows :
https://www.inaturalist.org/identifications?current=false&taxon_id=47792

NB : taxon_id=47792 = Odonata . The taxon_id which should apply to the community accepted ID correction, recognized by the “improving” label (above, Libellula quadrimaculata).

NB this topic is similar to this closed one that found no obvious solution.

Why : because “improving”-type “disagree” IDs are likely a way to recognize stronger expertise among identifiers (cf the several topics on experts in this Forum).

Thank you.

Looks like you want a combination of:
https://www.inaturalist.org/observations/identify?taxon_id=47792&identifications=most_disagree

and

https://www.inaturalist.org/identifications?taxon_id=47792&category=improving

Note that in the first, a lot of the options are improving. Many are also leading, but it’s not too hard to quickly pan through them.

How do you want to use this? To determine who’s the best identifier? If you’re looking for disagreements that are probably correct but need support, they would be leading. They would only be improving after the community agrees with them. (One exception would be IDs that disagree with finer IDs.)

3 Likes

Thank you. I did not know that parameter, but it does not seem to address my question.

The most reliable identifier(s) for a given taxon in a given region.

“Improving” is much more relevant than “leading”, though in practice the latter may be the only possibility for taxa and regions with small identifier communities (as discussed in other topics). Another problem with “leading” is that it is much more prone to gamification.

Yet this does not seem to be the solution. Indeed when I add the user_id option (e.g.) the “user_id” clearly is the observer of the observation containing a disagreement, whereas I would like to retrieve the disagreeing identifier. (and there will only be one per observation in the improving category).

Furthermore, it seems to me that many observations containing disagreements are missed from the search (by adding the respective observers as user_id=).

I note several (closed) topics discussing such “most disagree” and related parameters, yet for the quite different purpose of retrieving observations that might need further ID revision. These topics include 566 and 6698. In one more topic, I read that “these params operate on a notion of ‘disagreement’ that predates our current explicit disagreement model, so they don’t really mean much at all”.

Does ident_user_id fix that? Adding a modification to your above link gives all the observations in that taxon where that user has an active ID on an observation with a disagreement. However, that is not limited to improving IDs so I’m not sure how helpful that is for you.

You can read more about that parameter here:
https://forum.inaturalist.org/t/how-to-use-inaturalists-search-urls-wiki-part-2/18792

That’s sounds like a very tricky quality to quantify and I’m not sure this is the best way to measure that. But that’s starting to get into the philosophy of things that I probably don’t have enough of an opinion on.

Having an active ID in an observation with a disagreement is not the same as being the user who first disagreed. And even so, like before, the number of hits here is obviously far below the expected number, as all the many observations I know in this case are missing

.

There should be a “field” in the informatic structure of iNat observations that corresponds to these “disagrees this is” ID labels, just because they are stereotyped and appear as prominent banners visually.
I would have thought this “informatic field” as something easy to parse, just like the ID categories (improving, leading etc). And that combining it with the category and the user “fields” would be simple as well…, though perhaps not implemented as yet.

Well it’s returning all the observations that it should. It sounds like you’re also looking for observations where there was a disagreement, and the ID is still labelled as such, but the disagreement no longer exists because the original ID was withdrawn.

So you’re looking for active and former disagreements, which would all have the “disagree label”. The problem is that not all disagreements have this label. Things can get weird when there are taxon swaps and withdrawn IDs. Here is one example that still seems to be happening:
https://forum.inaturalist.org/t/disagreements-with-myself-are-not-labelled-as-disagreements/19429

If you’re looking for the most reliable IDers, why do you need disagreements instead of just improving IDs? That is already accessible:
https://www.inaturalist.org/identifications?taxon_id=47792&user_id=2520192&category=improving

1 Like

This would be very useful to find things that are very commonly misidentified

1 Like

You may be able to find some of that info on the Similar Species tab.

2 Likes

Yes, that must be the likely explanation for those very few hits, thank you very much for realizing it (because I was quite at a loss).

I will read the topic you mention, though it looks quite hard to follow at first sight ! In any case, my proposal does require that the “disagree label” remains the same, whatever happens to the ID it disagrees with (whether it is withdrawn, deleted, or changed back and forth…).

Why and by what mechanism should it change ?!
Only the category should be susceptible to change with changing IDs (from improving to leading or maverick etc).

Improving IDs are easily obtained by any middle-level expert, provided that there is a continuously large number of new observations by users who are less or not expert at all, as the latter usually do not venture into providing very specific IDs. This makes it simple for middle-level experts to easily accumulate improving IDs. .

Now there is a trade-off between accumulating easy IDs and trying to address difficult IDs. So middle-level experts get carried away in their happy run…, ending up with quite many mistakes. It is for the top experts to do the cleaning…, which starts by disagreeing and, if justified, associates with the improving label.

I mean that the number of “easy” improving IDs is overwhelming relative to the “hard” improving IDs, yet the latter need to be emphasized. The disagree-label seems a simple way to do this.

And it is not necessarily a matter of being top vs middle experts, but of being more or less attentive and careful. Yet it does seem to me that being top goes with being attentive…

Or you might ask a top expert, but first you need (a) good way(s) to identify her/him !

They don’t change, they just might not show up in the first place.

Every so often users disagree incorrectly. Not as common as a correct disagreement probably, but happens with enough frequency to make me wonder how accurate this assessment would be.

As long as the focus is on “disagreeing & improving”, other kinds of disagreeing should not matter.

i just want to jump in here for a moment and say that looking for observations is different from looking for identifications. i don’t think what odole is looking for in identifications can ever be quite approximated by looking for observations.

i also want to point out a couple of things…

first, for odole’s purposes, i think filtering on improving and disagreeing might be less effective than just filtering on disagreeing (assuming that filter is implemented at some point).

also, regarding thomaseverest’s suggestion for filtering by the identifications parameter in an observation search, it may not do what you’re actually expecting it to do. when you’re filtering using the identifications parameter, i believe the agreement / disagreement is with the community taxon, not with the other ids in the observation. so if you were thinking the disagreement was based on the latter, not the former, then might need to re-evaluate the usefulness of that filter. here’s some discussion on that topic: https://forum.inaturalist.org/t/how-to-exclude-observations-with-disagreeing-ids/15016/11.

that’s all i wanted to contribute here… hopefully this clarifies things a little rather than the opposite.

2 Likes

Thank you. I mean that one can disagree and be wrong; one could disagree by inadvertently hitting the wrong, unintended taxon name; and one could disagree for the fun of it (?!) etc. How would you exclude these “disagreeing” IDs that are obviously not relevant if you do not add the condition “and improving” ?

This is what I thought ; it is needed for the purposes of this request (see the argument just above).

This is what I suspected when I initially suggested URLs dealing with identifications :

And I just found something similar though not identical to the former, from the “How to use iNaturalist’s Search URLs - wiki” tutorial :

I will look into it…

Indeed, I think this is it, great :

https://jumear.github.io/stirfry/iNatAPIv1_identifications.html?disagreement=true&category=improving

One then needs to add the taxon_id and a user_id, and place_id if needed, and other parameters of interest (notably hrank=species should exclude the situations where the disagreement is associated with a coarser ID, which may be chosen by the community just because it is easier to agree with).

So, I am glad that the solution was simple. This tool (special thanks to @pisum !) enables us to compare the number of “disagreeing and improving” IDs with other parameters, and notably to the total number of IDs.

One expects that “disagreeing and improving” IDs will increase with the total number of IDs, but some users will show deviations above or below the expected number (that is an apparently faster or slower rate of increase than others), and this, I think, is a measure of expertise and/or … carefulness.
(And this should be visually clear in standard biplots and measured by residuals from standard stastitical regression).

In that format one can also retrieve and check observations individually and perform other types of parsing as suggested by @atronoxychump (and only partly addressed, in a black-box way, by the Similar Species tab).

Again thank you all very much. Thank you @thomaseverest for your interest and pedagogy !

You can also throw those first two links into the same url… https://www.inaturalist.org/observations/identify?identifications=most_disagree&category=improving

Not the exact same result as the github link, but it uses the ID modal which makes it really easy to use.

1 Like

Thank you. I tried it but it has the drawback mentioned above, which likely applies to any search through the iNat observations page.
If I add the option user_id, I get the observer of the observation that contains a disagreement and if I add the paramenter user_ident_id, I get any identifier associated with such an observation, whereas the purpose of this request is to retrieve a specific identifier and only that one: the identifier that raises the disagreement (and there will be only one disagreeing identifier per observation if the ID is of the improving category).

@thomaseverest @odole Sorry, for being unclear, what i meant was this function would make it easier to find the exact observations (not taxa!) where people make such disagreements so that others can independently decide what it is and change the community ID quickly

1 Like