Skip to content

Latest commit

 

History

History
33 lines (20 loc) · 1.52 KB

README.md

File metadata and controls

33 lines (20 loc) · 1.52 KB

Cross-lingual Medical Entity Linking Recommenders for INCEpTION

Simple integration of pre-configured xMEN pipelines as recommenders for (biomedical) entity linking in INCEpTION.

External Recommender

German SNOMED CT Quickstart (with Docker)

  • Download pre-computed xMEN index for SNOMED CT and extract contents into xmen_index:

  • docker run -m=12g -p <PORT_NUMBER>:5000 -v "$(pwd)"/xmen_index/index:/index/ ghcr.io/hpi-dhc/xmen-inception-recommender:main

Replace <PORT_NUMBER> with the port you want to use for the external recommender in INCEpTION. It will be available on http://localhost:<PORT_NUMBER>/xmen_snomed

Create xMEN KB and Index from Scratch

  • Adapt snomed_german.yaml as needed
  • Create KB and indices (takes about 1 hour with a GPU)
    • git clone https://github.com/hpi-dhc/xmen
    • xmen dict snomed_german.yaml --code xmen/examples/dicts/umls_source.py
    • xmen index snomed_german.yaml --all --overwrite

Running the Recommender without Docker

Install dependencies

  • pip install -r requirements.txt (see here for known issues during installation of xmen)

Run the recommender

  • Start the Ariadne Server (on http://localhost:5000/xmen_snomed):
    • python run_snomed_german_recommender.py ./xmen_index/index