MinimEL: Minimalist Entity Linking

The minimel package provides a framework to create and evaluate small Entity Linking models.

Warning This package is still under construction. A release is planned for the summer of 2024.

App

To run the app, run cd app and then flask run.

Evaluation Datasets

• VoxEL: A Benchmark Dataset for Multilingual Entity Linking

• Entity Linking in 100 Languages

• Tsai & Roth 2016

IDEAS

• per surfaceform, ignore entities that are an instanceOf the top entity

• NER features

• global binary classifier with (ent feat, sent feat) tuples

Tokenization, Stemming & Lemmatization

• Multi: https://github.com/mingruimingrui/ICU-tokenizer

• Multi: https://pypi.org/project/snowballstemmer/

• Japanese: https://github.com/SamuraiT/tinysegmenter

• Persian: https://github.com/htaghizadeh/PersianStemmer-Python

• Korean: https://pypi.org/project/soylemma/

Indices and tables

• Index

• Module Index

• Search Page