Biomedicus

What's Changed
* Downloads stanza models earlier by benknoll-umn in https://github.com/nlpie/biomedicus3/pull/117


**Full Changelog**: https://github.com/nlpie/biomedicus3/compare/v3.0.0-rc.3...v3.0.0-rc.4

Various bug fixes (Mainly RTF). Made it easier to programmatically create pipelines and use custom input sources. Better logging messages around configuration.

Breaking change:
The ``biomedicus run`` command argument for the output directory has switched to an option ``-o``. The correct syntax is now

bash
biomedicus run /path_to/input_dir/ -o /path_to/output_dir/


This is because the option is only used in the case of a serializer (default: json) and if the serializer is set to none the option will not be used.

This update contains the following enhancements for BioMedICUS

- Better negation detection via an adapted implementation of parsing-sensitive DEEPEN algorithm.
- Improvements to concept detection and vocabularies updated to UMLS 2020AA.
- BioMedICUS 2.x simple rule-based regex patterns section header detection.
- Fixed a bug which could potentially cause time-outs during deployment.

The fifth beta release for biomedicus 3.0.

- Negation via NegEx algorithm.
- Fixed an issue with normalized term concept detection.
- Added --host parameters to both "biomedicus deploy" and "biomedicus run". To host on the public IP use "biomedicus deploy 0.0.0.0".
- Fixed an issue where downloading would fail if ~/.biomedicus/data existed without a version file.
- Added dependency parsing via Stanford Stanza.
- Can execute java applications against a classpath containing the biomedicus Java jar by using "biomedicus java -- edu.umn.biomedicus.rtf.RtfProcessor --events localhost:50100 --port 50101" for example.
- Various other bug fixes.

Third Beta release for BioMedICUS 3.0.

- Fixed an issue that caused the sentences module to fail on unknown characters.
- Wrapper script which enables launching java against the included biomedicus jar.
- Initial baseline negation module provided by NegEx.

To install:

pip install biomedicus==3.0b3\[torch]

Installation instructions are available on our [website](https://nlpie.github.io/biomedicus/installation).

More-comprehensive models are available from https://athena.ahc.umn.edu/biomedicus-downloads/ for verified UMLS licensees.

The second beta release for biomedicus 3.0.

Includes the multithreaded pipeline runner from MTAP.