improving / replacing Identotron

[Ken-ichi]

How would people feel about replacing the Identotron with the
Suggestions tab in Identify? It's almost identical, except Suggestions
includes taxa from observations, checklists, and visual similarity,
not just checklists, but currently it lacks the map in the results
list and doesn't allow filtering by color or establishment means. We
could add maps to the results (probably), and est. means filtering
would work for checklist suggestions, but they're both a bit
problematic (loading lots of maps on a page is slow, est. means
filtering won't work with visual similarity). So, questions:

1) Are you using those maps in Identotron results? Is it enough that
Suggestions have maps when you click through the species details or do
you really want that map in the list of results?

2) Are you using the establishment means filter?

Personally I like those maps. I use the est. means filter like a few
times a year but could live without it.

[Charlie Hohn]

I like the suggestions tab because it includes the option to use the ID algorithm, checklists, or lists including non research grade observations. However, i definitely think the map is important, so you can see if something is even in range. I know I'm a map guy, but i vote for keeping it there. I guess if you get rid of it, I can click through, like you said.

[bouteloua]

1) Are you using those maps in Identotron results? 

No, but only because I don't use Identotron. I do use the Suggestions tab.

Is it enough that Suggestions have maps when you click through the species details or do you really want that map in the list of results? 

It would be nice to have maps on the same page, but adding visual similarity to Identodron would be a big enough improvement that I think it is worth it to implement it without maps. Perhaps when the website can handle it they can be built back in later.

2) Are you using the establishment means filter? 

Never. I rarely check this/search for this anywhere on the site, except to fix it when someone points out it is incorrect.

--

I would prefer if Identotron/Suggestions tab defaulted to a slightly higher Place and Taxon. I find myself changing it to a higher level almost every single time I use it.

e.g. state rather than county, family rather than genus

[Charlie Hohn]

absolutely agreed re: place setting for Identotron. it often defaults to a tiny park or other unit. I think anything smaller than the general range of county scale is cutting out tons of species for identotron, except where there are species lists. So if possible i'd say it should weight to anywhere with a full list, then to a county-type level or higher. For Vermont I always go to the state level even though we have high numbers of observations. But we are a geographically small state and in California I use county.

Also is there a way to filter out the options that lack geometry and don't work? When using identotron/suggestions i often have to try several places before i get one that queries properly.  Ideally those non-geographic ones would be removed from the site completely, but if not, at least removing them from here would be good.

[Ken-ichi Ueda]

Thanks for the replies so far. To some of your questions, the Suggestions tab already filters out "the options that lack geometry and don't work" (the Identotron doesn't, I think). And while we do have a concept of a "comprehensive" list it isn't used very much b/c it's hard to make a comprehensive list (some would argue that most of our checklists are shot to hell anyway and that it makes more sense to query observations directly in most parts of the world). Regarding what level in the hierarchy we choose for places and taxa, there's no right answer here so I don't think we'll change the current regime on the Suggestions tab. The right level for each is the one that *you* feel shows you a decent amount of data. In California, that's often the county since we have huge counties, most of which have lots of observations. In Vermont the counties are comparatively tiny and state makes more sense. If you live in Ireland, country probably makes more sense. And all these assumptions change based on what taxa you're looking at and what you know about the underlying data. If I'm identifying an assassin bug in California, I'll want state- and family-level results b/c I know those SoCal folks have done a better job documenting and IDing their bugs so I want to bring in their data. If I'm IDing an oak in the Bay Area I want genus- and county-level results b/c I know oaks are pretty well-covered at the county level in this region. While the current regime of just choosing the next rank up in the hierarchy doesn't work for everyone, at least it's consistent and predictable.

[Charlie Hohn]

yeah that all makes sense. Except i'd be surprised if many people really want to zoom in to a small park or town rather than at least county level. But it isn't that hard to zoom back out! and will get a lot easier if the geometry-less places are gone!

[Zanskar]

Hi there !

Great tool for me ! nice to have the suggestion tab inside Identify. Good job, thanks a lot !

[Matthew Muir]

I use Identotron frequently to browse what has been observed nearby, and like the maps even if they are slow to load and sometimes a little frustrating. I like that it's in the order of most to least frequently observed. 

My suggestions for potential improvements -- (1) being able to more easily compare photos of nearby observations of that taxon to my observation (instead of just the profile images). The profile images are less useful when there's geographic variability. (2) being able to use the annotations, so I can browse/compare photos of caterpillars vs caterpillars (not a mix of adults and caterpillars). (3) integrating the 'visually similar' and 'seen nearby' categories into Identotron, and greater transparency about what that means or how it's calculated. So, for example, if I look at Identotron for a Libellulidae observed in the DC Metro Area, common whitetails would be the top choice for "Seen nearby" with 18% of all Libellulidae observations in that Place = P lydia, and a % visually similarity or however iNat calculates that score.

I don't think I understand what you mean by "Suggestions tab in Identify" -- could be I've never seen that particular feature. 

On Thursday, September 21, 2017 at 11:44:34 AM UTC-4, Ken-ichi Ueda wrote:

[Scott Loarie]

To get to the "Suggestions tab in Identify"', navigate to the Identify tool, e.g. https://www.inaturalist.org/observations/identify?verifiable=true&page=1&taxon_id=47792&preferred_place_id=&locale=en-US&user_id= and click on an obs to bring up the modal. You should see Suggestions next to the Info tab.

Here's an old tutorial on how to use the Identify tool from BEFORE these tabs were added, but one that at least gives an overview of the tab:

https://www.inaturalist.org/pages/video+tutorials#identify

And here's a tutorial on the Annotations tab which is located next to the Suggestions tab

https://www.inaturalist.org/pages/annotations

We should probably make a tutorial specific to the Suggestions tab...

--
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+unsubscribe@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.

--

--------------------------------------------------
Scott R. Loarie, Ph.D.
Co-director, iNaturalist.org
California Academy of Sciences
55 Music Concourse Dr
San Francisco, CA 94118
--------------------------------------------------

[AfriBats]

1) Are you using those maps in Identotron results? Is it enough that
Suggestions have maps when you click through the species details or do
you really want that map in the list of results?

 I use these maps a LOT, and with various combinations of the overlays (ranges, GBIF records). I would be probably doing a fraction of my current IDs if I wouldn't have the geographic background information.

 

2) Are you using the establishment means filter?

No.

And I'd prefer if the choice of order by frequency vs taxonomy would be sticky - I prefer to see the list of suggestions without a bias what is supposedly the most common species etc.

[Russell P]

I do find myself using the maps. If one thinks of this as a field guide (which is how I think we're using it--only not while in the field), then I think maps should be readily available. Getting to them via the suggestions tab is easy enough, but then one has to zoom out to see the relevant distribution. If the maps can't be in immediate view, can they at least be made to appear more like how they appear in Identotron--already zoomed out at some relevant scale?

I never used the establishment means filter.

Also, can you help me understand how the filters work in the Suggestions tab? "Places" in particular. I feel like I need to know how this is working before I can use the Suggestions tab in any meaningful way. For example, for this observation: https://www.inaturalist.org/observations/9806360

https://www.inaturalist.org/observations/identify?quality_grade=needs_id%2Cresearch&order_by=random&radius=500&lat=32.2150262&lng=-98.2533024&taxon_id=107227

In "Identify" mode, it shows *marine* species if one selects "Arkansas" and "ray-finned fishes" and "visually similar". And there are no marine species in Arkansas. What's the value of filtering by place if it's not filtering by place?  Can there be a pop-up help button that explains things like this?

[Jane Widness]

It looks like the default zoom level for the Identotron was made the country level -- thanks, it's a major improvement over the previous tiny zoom default.

But can I request (if possible) that it move down to something like state/province/territory for the biggest countries, i.e. Russia, China, the US, Canada, Brazil, and Australia?

[Star Donovan]

Agreed.  It's very frustrating to look at Identotron to get some ideas for local taxons and have 16 pages of results because it is looking at the entire US instead of just the Austin Metro Area, the TPWD wildlife district, or even the state.

P.S. re: Ken's questions

1) Are you using those maps in Identotron results?  Yes.
2) Are you using the establishment means filter? I was not aware of this feature.

[Charlie Hohn]

I think for large countries such as the US and Canada, country is too large. For California, county seems best especially since at least plants have full lists by county. For Vermont, which is much smaller and less spatially diverse, state is fine. For looking at the map the towns are great, but for identotron they are nonsensical because they are so small. For Quebec, which goes from maple forest all the way to tundra, some smaller division would be better too, though I am not sure which administrative divisions they most use there. For a smaller country, the whole country may b e fine. Depends on how diverse it is. For instance I imagine a place like Costa Rica that is super diverse with dry and wet areas, you'd want a few divisions.

[Star Donovan]

Oh, wait. I've seen this.
I rarely use the establishment filter in the Identotron, because I rarely know the status unless I already know the species.