search

AtlasOfLivingAustralia / ala-name-matching

Atlas name matching API and index generation
Other
10 stars 13 forks source link

Pterostylis sp. aff. boormanii (Sunset Country) being returned rather than Pterostylis sp. aff. boormanii #207

Open Sherrin-ALA opened 1 year ago

Sherrin-ALA commented 1 year ago

In the name matching Unit Tests - AlaNameSearcherTest.testAffLookup1 Pterostylis sp. aff. boormanii (Sunset Country) is being returned rather than Pterostylis sp. aff. boormanii.

(Sunset Country) is an excluded name, but is coming up with a higher score the Pterostylis sp. aff. boormanii,

Sherrin-ALA commented 1 year ago

Notes from Doug:

Worked it out, The taxon.txt entry is https://id.biodiversity.org.au/instance/apni/51441972 https://id.biodiversity.org.au/taxon/apni/51441976 ICBN Pterostylis sp. aff. boormanii heterotypicSynonym species dr5214 https://id.biodiversity.org.au/instance/apni/51441972 https://id.biodiversity.org.au/name/apni/194759 Plantae Orchidaceae Pterostylis sp. aff. boormanii Pterostylis sp. aff. boormanii https://id.biodiversity.org.au/reference/apni/51428473 CHAH (7 July 2021), Australian Plant Census Backhouse, G.N. & Jeanes, J.A. (1995), The Orchids of Victoria [273] 1995 https://id.biodiversity.org.au/instance/apni/51441972 Placeholder name has been treated as unique The important bit here is the Plantae Orchidaceae bit. This will place this in familty Orchidaceae and kingdom Plantae. So the base score is 6000 not 5000 then -1000 and -200 for 4800. The plain usage is a synonym and doesn’t get the 6000.

The core problem here is that synonyms don’t have much higher taxonomic information and so don’t get the “owned=by” boost. Fix would be to either have a synonym inherit it’s defaultScore from the default score of the accepted taxon, as well as the parent for accepted taxa (not a bad idea) or adding an extra rule to detect “Genus (sp.?)? (aff.?)? (cf.)? epithet (Geographic)” in the name (hard to do without complications with authors) I've done some experiments with the synonym approach for baseScore and it's going into some infinite loops ... Thanks NZOR

Sherrin-ALA commented 1 year ago

An additional issue here may be that Pterostylis sp. aff. boormanii (Sunset Country) isn't being attached to the genus Pterostylis, it's being attached (as a species) to the family Orchidaceae - even though in the taxon.csv file from APC, it's ParentNameUseageID is identical to Pterostylis boormanii which is being placed properly under the genus.

Need to figure out why that's occurring - this may be causing excluded names to be matched to a family level rather than a genus level.