Smarter ranking of taxon search results during identification

During identifications as an identifier I would like to get the most relevant hits when typing a name or its abbreviation, so that I do not have to browse through a lot of different, but irrelevant hits, which when selecting erroneusly will lead to erroneous IDs.

Case 1:

In this case I tried to further identify the fly that was identified earlier as Chrysopilius sp to Chryspilus erythrophthalmus. I typed in Chr ery. It seems that this combination matched with several species, sometimes quite surprising ones like the first one Glacier Lily. It seems that the search phrase works with the entire phrase including the traditional names.

Having chatted with the inat people it seems that the hits to any search phrases are returned in an order that contains the most common hits (most observations for the taxon), rather than the ones which are relevant in the existing identification context. Now, something is already identified to some extent, the typical use case is that this identification will be further specified. Therefore, I would like to receive the hits which are within the taxon context first. I.e. I would like to see Chrysopilus erythrophthalmus as hit nr 1., and I would like to see all other hits that are outside the taxon context lower, and possibly marked somehow at the UI level as being a potential “maverick”. Several ways can be used to achieve this goal.

While the small photo icons help, in most searches the correct hit comes first, and when identifying hundreds of observations, one tends to hit always the first hit. This regularly leads to misidentifications.

See for example:

A fly was identifeid as Bombylius major. In fact it is Bombylius aurifer. In this case I typed B aurifer, and the clicked on the first hit, which turned out to be a snake. This happens quite often.

A similar one:

Here a Rhagio needs to be identified. Rha atl brings Rhagio atlanticus not as a first hit.

Another quite typical example, which happened not only to me, but others as well. Lomatia seems to be a name of a plant as well as a fly genus. The plant genus always comes up as the first hit, as this seems to be more observed than the fly.

There are countless other examples.

i think this is smart. the computer vision suggestions seem to do something similar, where they are limited by existing iconic taxon of the observation. so then in the case of the taxon autocomplete suggestions, it would be nice if taxa in the same iconic taxa – or maybe in the same taxon ancestry chain – came up first in the list. for example, if it’s already Rudbeckia, it would be nice to be able to just type “r hi” to get all the Rudbeckia hirta suggestions at the top.

i think it might be nice to have a visual (icon) indicator of the iconic taxa on the autocomplete list, too. it’s possible something like that would make things too busy visually, but on the other hand, it might help to mitigate situations where people select, say, a similarly named plant when they mean to pick an animal.


I think a change like this would probably help reduce misIDs in the long run, and I would support this. I do think that it might lead to a little confusion/some initial misIDs if first enacted however. I’d guess a lot of people like me have their most common IDs and search strings and the order of results memorized (and in muscle memory). So if this were changed, I think it would be good to publicize widely ahead of time, so people could be on the lookout for changes to their usual results.

Also, probably just a good reminder that we should all look carefully at our IDs before entering and not assume the system will behave exactly how we expect!


That is where I would like to be able to feed back a down vote.
NO. This is NOT a snake.

Would ranking by iconic taxon (or similarly coarse taxon) work for most cases? I feel that should be enough.

rank by prioritizing taxa with the same iconic taxon first. is this enough?

  • yes
  • no

0 voters