Evaluate how difficult feature parity with MetamapLite is

allenai / scispacy

A full spaCy pipeline and models for scientific/biomedical documents.

https://allenai.github.io/scispacy/

Apache License 2.0

1.71k stars 229 forks source link

Evaluate how difficult feature parity with MetamapLite is #66

Closed DeNeutoy closed 2 years ago

DeNeutoy commented 5 years ago

[ ] Term normalisation
[ ] Entity Linking
[ ] Negation Detection

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6080672/

kaushikacharya commented 4 years ago

@DeNeutoy In the Conclusion section of your paper, your team has mentioned negation detection as a future work. Has the work on "Negation Detection" feature started?

As per my knowledge, these are the following algorithms for negation detection in clinical notes:

MetaMap/ MetaMapLite uses variation of NegEx.
NegBio uses patterns in dependency parsing.

Is anything similar planned for negation detection in scispacy? Or are you planning a different approach?

dakinggg commented 4 years ago

Someone else created a spacy pipe based on the NegEx algorith, here: https://github.com/jenojp/negspacy. I haven't tried it, but you should be able to just use it with scispacy.

MichalMalyska commented 2 years ago

@kaushikacharya @dakinggg I think the best resource for that would be medspacy: https://github.com/medspacy/medspacy It's a step-up from just NegEx