Release Notes: v2.5.0 — Qualitative-Coding Wave (IAA Kappas, Codebook & Quotation Exports, Code Analytics)¶
Adds three Tier-1 capabilities that make Potato credible for qualitative-research workflows alongside its existing NLP/ML annotation surface. 32 new unit tests, all passing.
Inter-Annotator Agreement¶
- Cohen's kappa (pairwise, averaged over rater pairs) and Fleiss' kappa (N raters) in
potato/agreement.py, surfaced via/admin/api/agreementalongside the existing Krippendorff's alpha. Reviewers in the QDA space almost universally ask for kappa specifically; this closes that gap.
New Exporters¶
codebook— per-schema CSV with code hierarchy, color, description, and use count.quotation_report— per-span CSV with text, offsets, source document, and coder.
Both are deliverables researchers expect, and both are registered in potato/export/registry.py.
Code Analytics¶
Two new admin analytics endpoints that turn the dashboard into a basic query surface over data Potato already has:
/admin/api/code_cooccurrence— pairwise code co-occurrence across instances, with a schema filter and a min-count threshold./admin/api/code_crosstab— codes-by-instance-attribute pivot using existing instance metadata.
Bug Fixes¶
- Fixed
compute_config_md5crash on mixed-type dict keys (#153).
Upgrade¶
pip install --upgrade potato-annotation==2.5.0