Datasets:

---

brennercruvinel

/

3char

Like 0

Every record now has a label, and the six systems that were held back are published with their license_status unchanged: 64 systems and 817,340 rows, with no empty code or label.

• bio/pdb, trp/tld, cmp/git, fin/cus, cmp/ext and others take their names from official sources declared, pinned and checksummed in 3char-pipeline
• com/tlg is rebuilt: 15,672 code words with their phrases, replacing 7,359 rows in which the first word of the phrase often stood in for the code word
• wrt/tz abbreviations are those zic prints; grf/lpg and grf/gql keep only the languages' keywords
• retrieved_at in systems.parquet is corrected to the recorded retrieval dates
• figures/similarity_matrix.svg and figures/similarity_pairs.svg are added

README.md and CHANGELOG.md are not changed here; the card is updated separately. Until then, the six new systems are reachable through the all configuration only.

[POCS] Open data/bio/pdb.parquet, data/trp/tld.parquet or data/com/tlg.parquet in the file viewer of this PR and see the names filled in.

Built by https://github.com/brennercruvinel/3char-pipeline/pull/1