Release 0.9.1

Release 0.9.1 is available !

Changelog from 0.8.1

[Added/Changed]

• [text] attrs_to_copy for segmentation operations
• [text] store section names in Attribute rather than metadata
• Trainer logging changes
• [text] change default params for RegexpMatcherRule
• [text] add default rules for rule-based operations
• [text][preprocessing] Change Normalizer name
• Add e3c corpus support

[Fixed]

• [text] fix syntagma tokenizer rules
• [text] Harmonize kb_id field between EntityNormAttribute and other places
• [text] fix duplicate finder
• [umls-coder-normalizer] Don't crash on empty list of entities

[Config]

• Drop python 3.7 for bumping to torch 2.0
• [audio] remove pyannote-audio dependency from medkit package

[Docs]

• Add doc for load_text_documents
• [readme] add contact and link to CONTRIBUTING.md
• update medkit logo

You will find the documentation at: https://medkit.readthedocs.io/en/0.9.1

Finally, to install medkit, the package is available on PyPI under the name of medkit-lib

We strongly recommend to create virtual environments (using venv or conda) before installing medkit.

# Install medkit with only required dependencies
pip install medkit-lib

To install with extra dependencies (e.g., hf-translator, hf-entity-matcher, hf-transcriber, quick-umls, resampler, rush-sentence-tokenizer, umls-coder-normalizer, webrtc-voice-detector, ...):

# Install medkit with all extra optional dependencies
pip install 'medkit-lib[optional]'