Aurora Universities Network
Published November 8, 2022 | Version 1.1
Software Open

SDG BERT - Multi-language Multi-label BERT model for classifying texts to Sustainable Development Goals (SDGs) based on Aurora SDG Query Model v5

Creators

  • 1. Palacký University, Olomouc

Contributors

  • 1. Vrije Universiteit Amsterdam
  • 2. University of Southern Denmark

Description

SDG BERT is a Multi-language Multi-label BERT model for classifying scientific texts to the Sustainable Development Goals (SDGs).

The data used to train the model is based on Aurora SDG Query Model v5.

It's input a text fragment. Its output is an ordered sequence of 17 probabilities. The first one represents the probability to SDG 1, and so on.

 

 

 

 

Notes

Acknowledgements

Many thanks to Maéva Vignes from University of South Denmark, to allow us to use their UCloud HPC facilities and budget to train the mBERT models on their GPU's.


Funded by

Funded by European Commission, Project ID: 101004013, Call: EAC-A02-2019-1, Programme: EPLUS2020, DG/Agency: EACEA


Read more

[ Project website | Zenodo Community | Github ]


Change log

2022-08-31 | v1.1 | small tweaks to improve the accuracy-

2022-07-31 | v1.0 | first version of the multi-label model-

Files

Files (2.0 GB)

Additional details

Related works

Compiles
Dataset: 10.5281/zenodo.6644946 (DOI)
Is compiled by
Dataset: 10.5281/zenodo.5205672 (DOI)
Is documented by
Report: 10.5281/zenodo.5603019 (DOI)