This PR swaps ScispaCy for PubTator for pipeline-based entity hinting. This requires less code and dependencies (and is also significantly faster) but should lead to very similar performance.
Going to re-run all experiments and make sure results are as expected before merging.
Other changes
♻️ Drops the align and preprocess utils, moves them all into common/util.py
This PR swaps ScispaCy for PubTator for pipeline-based entity hinting. This requires less code and dependencies (and is also significantly faster) but should lead to very similar performance.
Going to re-run all experiments and make sure results are as expected before merging.
Other changes
alignandpreprocessutils, moves them all intocommon/util.pyCloses
Closes #34.