Implement more filtering tools for Casual grade observations

Formalizing my request from this post

Categories of the Casual section could be broken down usefully to add identifications (and/or to fix glitches) on Casual-grade records of interest. Preferred first choice to exclude from my perspective: Unknown place or date. Filtering might be achieved by extra filter checkboxes being added to the filtering pane of the Identify modal; or by letting flags’ status be queried via elements that would go into a search url.

-Unknown place and/or date [already doable, see below]
-marked date and/or location Not Accurate
-marked No/No recent evidence of organism
-marked Can’t be improved
-Unlabeled records by users who have opted out of Community ID
-Records with Copyrighted Media Removed

Note: “Cultivated” subset is already filterable, see

Update:“Unknown date” subset can already be removed, see

Update: Most of “Unknown location” subset can be removed? see

To look for Casual Unknowns that are not Captive and missing a lot of the broken elements, therefore use this chunk in a search url:


To search for the Captive side, just switch to captive=true to give:


BTW if adding searchability by flags, refinement by adding searchability by flag user may have its upsides in being able to find and fix sets made by a “vandal”, or by someone misunderstanding or repeatedly misclicking on a flag. (Downside of that though could be increasing chance of interpersonal conflict between people with strong opinions on a particular flag.)

1 Like

what do you mean by flag?

Each time someone ticks a DQA box like the ones in these examples, I think of that as a “flag”. But, I don’t know the terminology the site builders actually use, so feedback is welcome from that side.

-marked No/No recent evidence of organism
-marked Can’t be improved


Just crossposting some answers to these for easier findability in the future:

Casual Unknowns, excluding some broken elements, that are not captive

That are captive:

(looks like it’s just a couple hundred difference from the bounding box links above)


I have found that the geo=false or =true chunk still doesn’t respectively pull up or exclude a strange set that are captured by the bounding box urls I have above, but still have blank coordinates and report a “Location Unknown” map.

However, this tiny bounding box can fish them up. Here’s everything whether 0,0 or blank:

1 Like

I do refresh this conversation, as i liked to ask if there is a simple way to filter obss. which are missing (visible) location, thus won’t show up on maps for reviewal.
It’s tricky and much effort to find all obss. which lack (visible) locations, in especial in case of captive / cultivated ones.
As defined places got place_ids, i may imagine it shall be easy to sort a distinct place_id in case of missing location. Perhaps the number 0 was free for this purpose, and one needed just to type “missing” to get there.
Is a likewise feature already present, anyone who knows, please?
Why IDing casuals lacking places? Locations might be added later on, such reports could be of scientific value, e.g. associated with overlooked organisms, unknown ranges, new adventives, whatever.

I think geo=false will get you there (being sure to exclude obs with private locations, as well as including casual obs by unchecking the Verifiable box):,obscured&place_id=any&subview=table&taxon_geoprivacy=open,obscured&verifiable=any


Thanks a lot! This solution is nearby perfect, as it gives all reports lacking locations, except the ones which got, but were flagged for inaccuracy of location. Still these might perhaps be found by search for places, i will prove this.


I would like a filtering tool to either find or exclude all casual observations with community IDs (e.g. more than 2/3 agreeing identifications) from a search, preferably in the identify window. This is basically the “casual RG equivalent” and I think these are eligible for the CV training set. (If that is wrong, someone correct me please.) Is there a URL hack for this? I can’t seem to find anything.

The purpose of such a filter would be to more easily find observations that are potentially feeding into the CV training set but may have been misidentified. It seems once an observation is made casual, very few incorrect IDs are still getting reviewed and corrected. Part of the reason is probably that there is no way to narrow down the pool by doing the equivalent of sorting them into RG and Needs ID as for non-casual observations. Going through all of them just takes a lot of time and effort that could be more targeted to the observations where corrections will have a chance of making a difference for CV accuracy.


Update: @jeanphilippeb is successfully and helpfully breaking out chunks of Casual into accessible sets for review, eg,

I really appreciate the ability to go back through a DQA I know I’ve used a lot, and resurface some of the old ones (as I had alluded to in ) ! :)


About 68,000 observations got a comment “duplicate”:

About DQA flags: