Skip to content

CHANGELOG.md

Latest commit

 

History

History
80 lines (62 loc) · 5.56 KB

CHANGELOG.md

File metadata and controls

80 lines (62 loc) · 5.56 KB
Raw

CHANGELOG

Unreleased

Migration guide: See docs/migration-guide.md for step-by-step upgrade instructions.

Breaking Changes

  • evaluate_all() removed — raises DeprecationError at runtime. Replace with the three-step pipeline: predict_dataset() → CanonicalMapper.get_mapped_results_dataframe() → calculate_score_on_df().
  • entity_mapping parameter removed from SpanEvaluator, TokenEvaluator, and BaseEvaluator — entity mapping is now the responsibility of CanonicalMapper.
  • compare_by_io parameter removed from evaluator constructors — BIO/BILUO prefix stripping is now performed by CanonicalMapper.
  • BaseEvaluator.from_dataset() removed — use model.predict_dataset(dataset) directly.
  • Non-Presidio model wrappers removed: FlairModel, SpacyModel, StanzaModel, AzureAITextAnalyticsWrapper. Add models directly through Presidio to evaluate them.
  • Minimum Python version raised to 3.11 (was 3.10) — required by numpy >= 2.4.0.
  • Package manager changed from Poetry to uv — install with uv sync, run with uv run.

New Features

  • BaseModel.predict_dataset(dataset) — runs the model on a list of InputSample objects and returns a 5-column DataFrame (sentence_id, token, annotation, prediction, start_indices).
  • CanonicalMapper — replaces EntityMappingHelper with an improved four-tier auto-resolution strategy (EXACT, COUNTRY, FUZZY, PENDING). Key methods:
    • CanonicalMapper.from_dataset(dataset) — builds a mapper from dataset labels.
    • mapper.get_mapped_results_dataframe(results_df) — applies entity mapping to a predictions DataFrame.
    • mapper.get_mapping(mode='html' | 'text') — returns the final {raw_label: canonical | None} dict.
    • mapper.map({"LABEL": "CANONICAL"}) — manually resolve pending labels.
    • mapper.render_html() — display the resolution audit table in Jupyter.
  • TokenEvaluator.calculate_score_on_df(results_df) — score token-level predictions from a DataFrame.
  • SpanEvaluator.calculate_score_on_df(per_type, results_df) — score span-level predictions from a DataFrame.
  • Ruff — added as the project linter and formatter (ruff.toml at project root).
  • Pre-commit hooks — ruff format, ruff check, and pytest run automatically before every commit (.pre-commit-config.yaml).
  • Test reorganisation — tests are now grouped by topic (tests/data_generator/, tests/entity_mapping/, tests/evaluation/, tests/models/, tests/integration/). Integration tests are tagged with pytest.mark.integration.

Deprecations

  • evaluator.get_results_dataframe() — soft DeprecationWarning emitted at runtime. Replace with model.predict_dataset(dataset).

Version 0.2.5

Improvements

  • Introduced a new evaluator, SpanEvaluator which compares full spans of annotations and predictions, instead of tokens. (#141)
  • Make Azure SDK as an optional dependency (#116)
  • Add a DF output to evaluation results (#126)

Bug Fixes

  • Fixed bugs around plotting and experiment tracking (#140) around configuring Presidio in the evaluation loop. (#155)
  • Data generation bug fixes #113

Version 0.2.0

Breaking changes

  • Removed notebooks (pseudonomyzation)
  • Removed redundant classes FakerSpan, FakerSpanResult and updated code to use Span and InputSample respectively, changed SentenceFaker to inherit from Faker instead of using composition.
  • Removed functions from_faker_span, from_faker_spans_result convert_faker_spans from InputSample, as faker spans are now Spans so there no need for translation.
  • Removed PresidioDataGenerator to use PresidioSentenceFaker instead
  • Removed support for CRF models
  • Removed the FlairTrainer class, please refer to the official Flair documentation for training Flair models
  • Removed CRF as the package used is no longer maintained

Improvements

  • Improved evaluation notebooks: Notebook 4 shows a vanilla Presidio evaluation, notebook 5 shows a more customized Presidio with improved accuracy (#103)
  • Removed the Pseudonomyzation notebook as there is a more advanced approach within Presidio (#103)
  • Added the ability to use generic entities and skip words (#103)
  • Added the ability to do faster batch predict (#103)
  • Added sample_id to be able to reproduce the full sample (#103)
  • Fixed issue with hospital provider networking (#103)

Bug Fixes

  • Fix translation of Input Sample tags (#88)
  • Fix Presidio wrapper to call predict with a language parameter (#79)

Other Changes

  • Updates to all classes inheriting from BaseModel, as the predict signature has changed (now containing **kwargs) (#92)
  • Added Poetry instead of setup.py (#91)
  • Rename UsDriverLicenseProvider.driver_license to us_driver_license (#90)
  • Removed redundant classes FakerSpan, FakerSpanResult and updated code to use Span and InputSample respectively instead (#72)
  • Changed SentenceFaker to inherit from Faker instead of using composition (#72)
  • Simplified the use of SentenceFaker in the default option (RecordGenerator is instantiated if records are passed, otherwise a SpanGenerator is instantiated) (#72)
  • Updates to unit tests to support this change (#72)
  • Updates to poetry to include the config in setup.cfg, setup.py, and pytest.ini (#72)