• Table of contents
• 2013-05-30 conference call
  • Upcoming
  • To do for Martha
  • To do for Brad
  • To do for Aaron
    • load GBIF with current filtering rules
    • postprocess TNRS results to exclude animals with genus homonyms
    • fix TNRS speciesBinomialWithMorphospecies to include alternatives when no accepted name
    • fix higherPlantGroup to match on the genus when no family match
    • add COALESCE of TNRS accepted and matched name to analytical_stem_view
    • FIA filtering
  • Availability

2013-05-30 conference call¶

Upcoming¶

• there will be a call next week, but Brad will not be available

To do for Martha¶

• request that Aaron get access to the production TNRS DB Edwin at iPlant is providing access
  • this contains the Tropicos genus homonyms

To do for Brad¶

• e-mail out the algorithm for postprocessing TNRS results to exclude animals see Result filtering
  • see tentative algorithm below under "postprocess TNRS results to exclude animals with genus homonyms"
• the IRMNG database's *animal/plant genus homonyms* are available on the web, so probably no need to contact Tony at CSIRO to get them
• determine whether we also need to deal with animal/plant family homonyms
  no, "the only homonyms involving plants are plant-plant." "There no family+genus pairs shared by both animals and plants." (Brad)

To do for Aaron¶

load GBIF with current filtering rules¶

• we aren't changing the subsetting, because we will postprocess the TNRS results instead
• we aren't lowering the % plants threshold below 80%, because we want to avoid animals in the DB at all costs¹

postprocess TNRS results to exclude animals with genus homonyms¶

DONE: see Result filtering

• this will be done for all datasources, not just GBIF

1. DONE: if family is exact match (Family_score = 1 _²) and is plant family, assume plant and keep
   The Family_score check is necessary because TNRS transforms animal to plant families via the fuzzy matching
2. else if genus is in *IRMNG animal/plant genus homonyms*, assume ambiguous and discard
3. else determine animal/plant status based on higherPlantGroup done in a separate filtering step (not part of TNRS postprocessing, because higherPlantGroup is not available yet at that point)

fix TNRS speciesBinomialWithMorphospecies to include alternatives when no accepted name¶

• use accepted name, else matched name, else Name_submitted

fix higherPlantGroup to match on the genus when no family match¶

• create genus->higherPlantGroup lookup table
• lookup table must exclude internal plant homonyms (different from animal/plant homonyms)
  • get these from TNRS's Tropicos DB

add COALESCE of TNRS accepted and matched name to analytical_stem_view¶

FIA filtering¶

¹ "one animal in the range modeling could throw everything off" (Brad).
Note that BIEN2 did not filter out animals, causing problems in the range modeling that must be fixed in BIEN3.

² A match score < 1 always indicates a fuzzy match (Brad)
.

Availability¶

• Brad will be unreachable all of next week