Aurora Universities's repositories
sdg-queries
This repository contains machine readable (xml) search queries (crafted from a controlled vocabulary), for the Scopus publication database, to find domain specific research output that are related to the 17 Sustainable Development Goals (SDGs). We invite enyone to improve the SDG queries further in a co-creation process.
sdgs_many_berts
Classify Texts to SDGs in 100 languages. This repository shows the code used to train and predict multiple mBERT models; one per each SDG Target.
data-transform-scripts
script for enriching and adding data
sdg-queries-evaluation-report
In this study we evaluate the accuracy of our Aurora SDG classification model version 5, to match research papers to the Sustainable Development Goals (SDG's) of the United Nations. The aim of this investigation is to be transparent about the accuracy of the model, because this model might get used in reporting and strategy analysis by University leadership. The measurements are based on a baseline, 'golden set', where researchers of the Aurora universities handpicked publications that relate to an SDG. We measured the Precision and Recall of the Aurora model, and relating it to our previous version of the model, and to the Elsevier SDG model.
sdg_model_api
A small light-weight flask application to facilitate the deployment of various SDG classifiers.
sdg_query_execution
performing queries in scopus and enricment with additional data
SDGs_multilabel_classification
A script to process a dataset distilled out of a 50K samples dataset based on Aurora queries.