feat: add taxonomy classify service + /classify endpoint
Some checks failed
Deploy classify service / build-and-deploy (push) Failing after 24s
Deploy fil (kreuzberg) / deploy (push) Successful in 53s

- scripts/taxonomy.py: shared taxonomy with 14 categories, keyword scorer
  and classify_text() function
- scripts/classify_server.py: FastAPI service — forwards to kreuzberg /extract,
  applies taxonomy, returns category/subcategory/confidence alongside full kreuzberg response
- Dockerfile.classify: lightweight Python image for classify service
- classify.nomad: Nomad job → classify.i80.dk
- .gitea/workflows/classify.yml: CI/CD pipeline (build + deploy)
- analyse_familie.py: refactored to import from taxonomy.py (no duplication)
- .gitignore: exclude dokumenter_keywords.* and extract_all.log
This commit is contained in:
Henrik Jess Nielsen
2026-06-05 19:57:39 +02:00
parent f0300b586b
commit 58210207ea
7 changed files with 1027 additions and 0 deletions

6
.gitignore vendored
View File

@@ -245,6 +245,12 @@ obj/
pkg/
# Doc classifier output files
dokumenter_keywords.json
dokumenter_keywords.classified.csv
dokumenter_keywords.misplaced.csv
extract_all.log
# Local dev artifacts
docs/demo-dev.html
docs/serve.json