GBIF match results for concatenated file

[peterdesmet]

This table gives an overview of the current GBIF name matching status:

match	fishes	harmonia	macroinvertebrates	plants	rinse	rinse-annex-b	wrims	sum
exact match with gbifapi_scientificName			1	1294	36	1		1332
exact match with gbifapi_canonicalName	22	130	66	3	6093	21	175	6510
EXACT 100%		4		849	211	6		1070
EXACT < 100%				72	20	4		96
FUZZY	1		2	15	73	4		95
HIGHERRANK		3	4	177	224	7	25	440
NO OR DOUBLE MATCH		3			4	2		9
sum	23	140	73	2410	6661	45	200	9552

Some observations:

• The first 3 categories can be considered OK, which is 93,3% of the dataset! The only caveat is that we have to trust the accepted names GBIF gives for synonyms (745 records + 15 doubtful), which we don't always do: e.g. Tripolium pannonicum is not a synonym of A. salignus
• The 96 EXACT < 100% matches will have to be examined case by case.
• The 95 FUZZY matches are probably typos, and are addressed in #41
• The 440 HIGHERRANK matches are mostly plants and a lot of hybrids. Chances are we can only correct half of those to match in GBIF.
• And then there are 9 records with no match: those are some viruses and names in Harmonia that appear twice in GBIF (a bug that will be fixed in April).

@timadriaens, how do you want to prioritize going forward?

[peterdesmet]

Priority list:

1. Synonyms in first 3 categories: to be checked by Filip Verloven
2. Correct FUZZY
3. Verify EXACT < 100%
4. Correct NO or DOUBLE
5. HIGHERANK: correct Plants in collaboration with Filip
6. HIGHERANK: correct WRIMS in collaboration with VLIZ

[peterdesmet]

Update:

• Synonyms should be verified indeed (now highlighted as such in data)
• There are no FUZZY matches as we use strict mode
• We dropped the confidence column: GBIF now also matches on authornames + all matches were checked for rank to: we should be save there.
• NO or DOUBLE: no longer occurs
• HIGHERRANK: no longer occurs because of strict mode

[peterdesmet]

Will create separate issue for synonym verification and close this one.