Named Entity Service

Authors: Sören Räuchle, Nadav Babai, Jan Fillies | 3pc GmbH

This service is analyzing text for extracting Named Entities, Linked Entities, Abbreviations and Sentiment Scores for the German language.

Named Entities

The Named Entities are a hybrid of AI and Gazetter Dictionary matching. The AI is looking for the Grammar and is using a Spacy Model. The Gazetter Matching is based on Entities extracted from the 3pc.de Website.

Abbreviation Detection

The Abbreviation Detection is based an Algorithm described in a Paper of A.S. Schwartz and M.A. Hearst. Additionally, the Detection is extracting Abbreviations from HTML Content.

Sentiment Analysis

Sentiment Analysis is done with a Sentiment Dictionary for the german language in combination with a lemmatizer.

Install, Run, Test

• Install dependencies: conda env create -f environment.yml
• Install Testing dependencies: conda env update -f environment.yml
• Run: python __main__.py (inside the source code directory)
• Run Tests: pytest (inside the tests directory)
• Generate protos with: bash python generate-proto.py

Technology Stack

• Python 3.8
• FastAPI
• Conda
• Spacy
• More dependencies envionment.yml
• More runtime dependencies docker-compose.yml

To activate profile, add to env variables:

PROFILE=local

More Info | Documentation | Papers

• A Simple Algorithm for Identifying Abbreviation Definitions in Biomedical Text