Identify Suggestions on Obs Detail Page

We're running a test of using the Suggestions from the Identify tool on the observation detail pages on the web instead of the Identotron (click the button at the bottom of any obs detail page). On the plus side, you don't have to leave the page, you can look at options from observations and vision in addition to just checklists, and you get a little more info about each taxon. On the minus side, no more quirky name, can't peruse a lot of maps at once, no color filtering (not that that was really working). Mostly just posting here as a way to solicit bug reports.

Anyway, if nothing really bad turns up in a few days we'll make this the default. Holler if you find any problems.

Posted by kueda kueda, July 06, 2018 11:38 PM

Comments

Thumb

This is great, thanks! All the below would also apply to the Suggestions tab on Identify.

1) The zoom in button on the "Observations / Map" taxon maps overlaps with the "toggle fullscreen view" button for me on a 15" laptop (view).

2) Visually Similar does not appear to restrict it to place, even though a place is selected. For example, https://www.inaturalist.org/observations/14106639 there are several Spiraea species included when you select "visually similar" that do not have observations in St. Louis County, MN. I think the fix to this would be to clear out the place by default when visually similar is selected, and to actually apply the Place filter if it is re-selected manually. 2b) FWIW I also think the taxon should be cleared out by default when visually similar is selected.

3) There is a large/jarring gap between the taxon name and the "About" snippet from Wikipedia.

4) The number one reason I need to open up the full taxon page after browsing Suggestions is to view the taxonomy. Would be nice to squeeze that in underneath the taxon name in that big gap, or under the map or something.

Posted by bouteloua 4 months ago (Flag)
Thumb

When you click a species from Suggestions, can it be made so that the maps are zoomable with the roller wheel of the mouse? Like elsewhere on the site? And elsewhere on the site, some places you have to hit CTRL and some places you don't. If possible, please be consistent.

Posted by pfau_tarleton 4 months ago (Flag)
Thumb

hmm. the interface is way better. However, the loss of the maps makes this a LOT less useful to me. is there a way to get the maps back in there somehow? The 'places' you can filter by rarely coincide with ecoregion and being able to see say, if something mostly occurs in the Green Mountans or Champlain Valley, in wetlands or uplands, etc is super helpful. I hope you figure out a way to get the maps more prominent again.

Posted by charlie 4 months ago (Flag)
Thumb

Thanks all, will take all this into consideration.

@bouteloua, can you explain *why* you want to view the taxonomy?

@pfau_tarleton, I think consistent use of the scrollwheel to zoom maps would hurt more people than it would help. Generally you don't want it to get in the way of scrolling, and Suggestions are one area where you need to scroll. If we put a taxonomic tree beneath the map, for example, it would be hard to scroll to it if the map zoomed based on the scrollwheel.

@charlie, agreed, but a) it is technically difficult to show those maps for lots of taxa (you either load 500 maps or you have to load them as you scroll to them, which we didn't manage very well on Identotron either), and b) space is limited. Joelle and I went back and forth about how to squeeze the map into the list and ended up deciding the photos were a better use of space.

Posted by kueda 4 months ago (Flag)
Thumb

@kueda Cases where it seems to quite possibly be one of the listed suggestions but I'm not confident enough to select a particular one and would prefer to go the shared family, tribe, or other closest ancestor taxon.

Posted by bouteloua 4 months ago (Flag)
Thumb

@kueda hmm... i mean it seems a lot less valuable without maps, i'm going to have to end up clicking through everything to see the maps each time if we keep this new format, because location matters a whole lot to ID, at least with plants. (can we have maps but just for plants?) Obviously i've got a map bias. In terms of bandwidth/data that makes sense and we just have to work with what we have, but in terms of space, i'm definitely gonna say this is a move towards more 'pretty' and less 'pragmatic'. So i'm going to say i still prefer the old format. Not a huge deal though... I don't use it super often.

Posted by charlie 4 months ago (Flag)
Thumb

@kueda, on some maps, you hover over the map and can zoom with the scroll wheel. On other maps, you have to CTRL click while using the scroll wheel while hovering over the map. After clicking on a taxa from the Identify view, NEITHER method zooms the map. I find it hard to believe that having to remember which of these three methods works in each of these situations is the best approach. Especially since the entire browser window zooms if you get it wrong!

I have to admit, the maps were nice. But I think having more pictures per species is superior if one can't have both pictures and maps.

Posted by pfau_tarleton 4 months ago (Flag)
Thumb

Is this why the maps are loading so durn slow, site-wide?

Posted by ellen5 4 months ago (Flag)
Thumb

i dunno but that's been driving me crazy too. I think it's just scaling in general. Site is growing a little too fast to keep its features running optimally

Posted by charlie 4 months ago (Flag)
Thumb

Compare is identical to the Suggestions tab in Identify mode, so I'd recommend using the same term in both places (I like Compare more than Suggestions, and that term accurately describes what one is doing in both places).

I also agree with bouteloua's item 2. It's very misleading because one thinks you're searching the selected geographic area for visually similar things. It's quite a let-down when one finds out otherwise!

Posted by pfau_tarleton 4 months ago (Flag)
Thumb

For the suggested taxa, can it somehow be made to exclude the observation in question?

For example:
-Echinacea laevigata is not known from Minnesota.
-On https://www.inaturalist.org/observations/14418366, click Compare.
-Under default suggestions, Echinacea laevigata appears (Cook County, MN, US; genus Echinacea; source: observations).
-However it only appears there because it's referencing itself, as it's the only obs of that taxon in Cook County, MN.

I know I can click RG observations as a potential alternative, but it seems like the default suggestions should exclude the current taxon when it's otherwise never appeared in that place on iNaturalist.

Posted by bouteloua 4 months ago (Flag)
Thumb

1. Could we have a quick link to the taxon page of each suggestion? Currently the Identify tool (unlike the Identotron tool) makes the user click once to focus on a taxon and only then can one get a link to the taxon page. I've done a mock up here with a red arrow showing the sort of link I'd like to see: https://i.imgur.com/QVaD5O0.jpg
2. I've never understood why one has to make an ID before one can access the Identotron? Why can't the 'compare/suggestion' link to the Identotron be there before any ID is made? I often make an ID to genus just to access the Identotron where I know I'll immediately find the species I'm looking for.
3. As Charlie says, please keep the maps. What is good about having two systems on iNat is we can choose the one that suits our style best.

Posted by vynbos 4 months ago (Flag)
Thumb

I've looked into the test version of the compare / identify function on the observation detail page (previously called the Identotron). It seems that the map display (showing the observation location alongside other iNat records, species ranges and, optionally, GBIF records) will be completely dropped in the currently planned update.

PLEASE, PLEASE, PLEASE do not drop the maps - this is one of the most useful tools on iNat with regard to identifications as it allows to critically examine iNat records, previous IDs and a lot more, all in the context of neat simple maps. I for one would reduce my identification activities on iNat a lot, simply because I don't have the time to open the map views of relevant species in separate tabs, and then trying to find out where the observation sits with respect to the other records (iNat, GBIF, ranges).

I really hope this lovely tool can be saved!

From https://groups.google.com/forum/m/#!topic/inaturalist/i0me1UWL4X8

Posted by jakob 4 months ago (Flag)
Thumb

Can you make the Places dropdown only show level 1, 2, and 3 standard places by default, then still allow users to search for other places if they want? Most people don't need to see 8 iterations of eastern US/North America or "BIOL272 Project Area" https://www.inaturalist.org/observations/10899393

I also think the default zoom level is too zoomed in for how I typically use those range maps. I find myself scrolling out at least 3 or 4 levels every time I use them.

Posted by bouteloua 3 months ago (Flag)
Thumb

Regarding the complaint that "Visually Similar does not appear to restrict it to place", what exactly are you expecting? What's happening is the computer vision system accepts lat/lon coordinates, not a place like obs search or checklists do, and it uses those coordinates both to apply the "Seen Nearby" label and rank frequently observed taxa higher. So in order to get some kind of geographic customization working, Suggestions just uses the centroid of the chosen place as those coordinates. Alternatively, we could use the coordinates of the observation to provide the vision system with geographic info, then filter the results based on... taxa that have been observed in the chosen place, or taxa on the checklist of the chosen place. That seems kind of complicated, but closer to what it *sounds* like you folks want, yes? If so, do you want to filter the results by taxa observed in the chosen place or on the checklist of the chosen place?

Posted by kueda 3 months ago (Flag)
Thumb

With the current settings for what is displayed when I select Visually Similar, I would expect the results to be filtered by the Place that is displayed. I would also expect the results to change if I select a different Place.

Using https://www.inaturalist.org/observations/14106639 as the example from before, I click Compare, then change the Source to Visually Similar and the filters at the top say "St. Louis County, MN; Spiraea; Visually Similar." But many of the options presented: S. betulifolia, S. douglasii, S. stevenii, S. splendens don't occur on iNat in St. Louis County based on observations *or* checklists. If I change the place to somewhere else, the Visually Similar results never change.

I *think* checklists would be better than observations. If you do go with iNat observations instead, would also need to implement the exclusion-of-the-current observation detailed above with the Echinacea laevigata example.

Posted by bouteloua 3 months ago (Flag)
Thumb

Following up on Cassi, I don't think we're suggesting changing how the Visually Similar *algorithm* functions, but rather how the option is presented to us. For example, if the Place went blank when one selected Visually Similar, then we wouldn't have the expectation of function that Cassi describes. Or separate out the Visually Similar option in some other way so the user isn't confused as to how it's being implemented.

Posted by pfau_tarleton 3 months ago (Flag)
Thumb

Ok, I've tried to implement some of your requests on http://gorilla.inaturalist.org (username: preview, pw: 313phant). The kind of taxon filter Cassi described for the visual suggestions is a bit trickier than I had time for today so I just removed the place filter when you're viewing visual suggestions. Maps in the list are cool but definitely slow things down a bit. Anyway, if these changes look ok I'll try to merge them in next week.

Posted by kueda 3 months ago (Flag)
Thumb

hmm, yeah i agree you shouldn't be able to choose a place for the algorithm if it doesn't weight the place seperately. Centroid is hard because what about if I use 'north america' one time and ' Montpelier, vermont' the other? for a large place a centroid can't really work.

As for the maps, yeah they make it slower but they are integral to ID especially for things that DON'T filter from a list so yeah, totally worth it. WIthout them i will have to click through to the maps and waste the bandwidth anyway and it's also annoying and would cause me to use the feature a lot less

Posted by charlie 3 months ago (Flag)
Thumb

Looks really nice! Thanks for improving this feature. : )

Posted by bouteloua 3 months ago (Flag)
Thumb

Hi @kueda, I'm super happy to see that the maps have been retained in the new design, thanks a bunch!

When using "Checklist" as the source, I noticed that inactive taxa occassionally pop up , which seems to be a bug and can hopefully be supressed.

Posted by jakob 2 months ago (Flag)
Thumb

It seems to work very well. I still would like to be able to initialize the compare feature on an obs with no ID. Perhaps when you type in an ID, and a drop down list appears with the 'view' on the RHS, the option 'compare' could be just to the left of it.

Posted by vynbos 2 months ago (Flag)
Thumb

I'm really digging the new format!

Posted by pfau_tarleton 2 months ago (Flag)
Thumb

@vynbos there is a smaller "Compare" link to the right of the Community ID header. Check https://www.inaturalist.org/observations/8744692

Posted by bouteloua 2 months ago (Flag)
Thumb

Jakob, regarding checklists, that's probably b/c a checklist has an inactive taxon on it. Checklists... have a lot of problems, but if you see inactive taxa on one, you can probably remove them by executing the arcane sequence of rituals required to find the listed taxon page. If you think inactive taxa are showing up in situations where there are no such listed taxa, that sounds like a bug, but I'll need to know the URL and combination of filters you're using to replicate it.

Posted by kueda 2 months ago (Flag)
Thumb

@kueda that seems to be the problem - taxa becoming inactive after taxon changes, and rather than being removed from the checklists, they're still included in the checklist (albeit as inactive taxa).
For instance, if I use Checklists as the source, Gabon as the place, and Pteropodidae as the filter taxon for this observation
www.inaturalist.org/observations/16113513, a whole bunch of inactive taxa will show up, which correspond to
www.inaturalist.org/check_lists/7453-Gabon-Check-List?taxon=40782&taxonomic_status=all

What about systematically suppressing the display of inactive taxa in the Compare-view? I can't think of any reason why we would like to see them there. Even better, kick inactive taxa from the checklists if that's an option.

Posted by jakob 2 months ago (Flag)
Thumb

Re: Compare" link next to the Community ID. For some reason this uses different place selection to the 'Compare' used next to an ID. Take this obs for example, if you click "compare" next to the ID "Gladiolus virescens" the place offered is the local place "Cape Winelands' with the drop down list "Western Cape" or "South Africa" screenshot. However if you click on the 'compare' next to the "Community ID" it only offers you "Western Cape" with the drop down list blank screenshot. Can we make the 'compare' next to the 'Community ID' have the same (more useful) places?

Posted by vynbos 2 months ago (Flag)
Thumb

Jakob, this is just one of many problems with checklists, so I think that's out of scope for this particularly piece of work.

I'll fix the different places showing up for different links. Theoretically they should all show the standardized places by default and not "community places" like "Western Cape."

Posted by kueda 2 months ago (Flag)
Thumb

@kueda no option to block inactive taxa showing up on the compare view?

Posted by jakob 2 months ago (Flag)
Thumb

@kueda thanks, the places are now working properly as requested.
Another request that I'm sure everyone would use extensively: would it be possible to let us utilize the 'compare' feature from the upload page? I'm thinking that once a user has made a tentative ID for all of their obs, prior to submiting the observations, one can check each of our IDs/narrow them down with the compare tool.

Posted by vynbos 2 months ago (Flag)
Thumb

I know this is beyond the scope of this discussion, but I'm currently trying to ID a narrow-waisted wasp which falls under Apocrita but that suborder include all the ants, so the suggestion box is dominated by ants, which it clearly isn't. My request would be that below the taxon dropdown list one had an 'without taxon:' dropdown list where one could exclude 'Formicidae'.

Posted by vynbos 2 months ago (Flag)

Add a Comment

Sign In or Sign Up to add comments

Is this inappropriate, spam, or offensive? Add a Flag