provide NER output from SQLlite/basex tables

[realmarcin]

NER methods were run on an earlier version of the Biosample data, specifically for the Title and Description fields. This process used a simple TSV file with these fields source from the Biosample table:

biosample id
title
description

For the next iteration, it should be possible to have an additional piece of code as part of the NER pipeline to execute a basex query and output the results in the TSV format above for NER input.

@hrshdhgd @turbomam

[turbomam]

NER/annotation is now part of EMA's annotation pipeline using OAK:

• OAK-based text annotation runs over env triad fields and attribute values
• Annotation sources: OAK, OLS API, BioPortal API
• Output: env_triads_flattened with provenance for each annotation source
• Makefiles/env_triads.Makefile — full annotation pipeline

[turbomam]

Reopening — my earlier close was inaccurate.

What EMA provides: OAK/OLS annotation output for env triad fields only.

The gap: NER output over broader free-text fields (Title, Description) from the SQLite/BaseX-era pipeline was never reproduced in EMA. The original issue described NER that ran on those specific fields — that scope is not covered.

The SQLite/BaseX infrastructure itself is gone, but the goal of having NER output over all textual biosample fields is still unmet.