run NER/CR over all textual metadata fields

[cmungall]

Execute RUNner over

• study table (see fetch study metadata #1)
• sample table

Run over all textual fields, in particular:

• description fields
• envo triad fiels

Vocabularies: ENVO, CHEBI, NCBITaxon in text fields, specifically

this can then be used to repair the tsv to insert the correct identifier for the ENVO class; also for prediction

[hrshdhgd]

I have completed my first pass at running NER against the metadata. Note I have only used ENVO as a dictionary for starters. We could add more ontologies as deemed necessary. The output can be downloaded from here.

@cmungall ; @wdduncan ; @realmarcin

[turbomam]

NER over textual metadata fields is implemented in EMA:

• OAK annotators run over env triad fields (env_broad_scale, env_local_scale, env_medium)
• Vocabularies: ENVO primary; OLS and BioPortal for additional coverage
• Output stored in env_triads_flattened collection with per-annotation provenance
• Makefiles/env_triads.Makefile

[turbomam]

Reopening — my earlier close overstated EMA's coverage.

What EMA does: OAK annotation runs on env_triad_component_labels only — the three env triad fields (env_broad_scale, env_local_scale, env_medium). This is confirmed by new_env_triad_oak_annotator.py line 152.

What's NOT done anywhere I'm aware of: NER over free-text fields like biosample Title, Description, or other non-triad attributes. The original issue asked for this broader coverage.

This remains a real gap. If anyone has done this work elsewhere (e.g. in INCATools, NMDC runtime, or a notebook), please link it here.