Flair NLP

We develop Flair, a very popular library for state-of-the-art NLP. It is used in thousands of industrial, academic and open source projects.

Tagging Text with Flair

Flair implements state-of-the-art approaches for various NLP tasks such as:

With a few lines of code, you can load one of our pre-trained models and apply it to your text!

For instance, our 18-class entity tagger can detect entities such as person names, dates, organizations, place names, and many others:

Flair supports tagging of text in many languages. In fact, many models are multilingual, allowing you to input text in any language.

Biomedical Models

Flair also includes the HunFlair family of models that allow you to tag and link biomedical text data. This allows you to detect the names of genes, diseases, chemicals and link them to normalized identifiers in a knowledge base:

For instance, in the above example, "autism" is detected as a disease name and linked to the term "Autism Disorder" in a standardized knowledge base. "Mice" is detected a a species name and linked to the entry "Mus Musculus".

Getting Started

Our tutorials explain how to apply NLP models to your data, and even how to train your own models. Since Flair was designed to be simple to use, the tutorials should be quick to complete.

Publications

Flair is my group's main vehicle for making our NLP research publicly available. Some important papers include:

Contextual String Embeddings for Sequence Labeling. Alan Akbik, Duncan Blythe and Roland Vollgraf. 27th International Conference on Computational Linguistics, COLING 2018.

FLERT: Document-Level Features for Named Entity Recognition. Alan Akbik and Stefan Schweter. arXiv 2020.

HunFlair2 in a cross-corpus evaluation of biomedical named entity recognition and normalization tools. Mario Sänger, Samuele Garda, Xing David Wang, Leon Weber-Genzel, Pia Droop, Benedikt Fuchs, Alan Akbik, Ulf Leser. Bioinformatics 2024.