The SDG-meter is a pilot artificial intelligence tool developed by UNEP and ISEP, Grande Ecole d'Ingenieur via a neural network-based technique for natural language processing (NLP) pre-training. It suggests interlinkages among user-submitted text and terminology from the 17 Sustainable Development Goals. While applications of this tool are being further explored, we departed from the practical use case where documents produced and consumed by UNEP needed to be mapped to the SDGs (project proposals, reports, briefings, etc.). Such mappings are time consuming and rely on personal knowledge of the links between topics and the SDGs. For more Information, please refer to the scientific paper.
This technique is based on the BERT (Bidirectional Encoder Representations from Transformers)
model developed by Google researchers (see paper here)
with the use of the multilabel text classification feature. Our algorithm has been trained with about 3000 texts and labels extracted
from the categories "News", "Guest articles" and "Policy Briefs" of the IISD-SDG website.
Our method has an accuracy of 98% on 500, which means that for 500 test texts our method correctly classifies 490 texts. To have an idea of all
the capacities of BERT and also its functioning explained in a very simple way we recommend you this short
video.
Contributors: Jade GUISIANO (ISEP-UNEP), Bharat BARADUR (ISEP), Raja CHIKY (ISEP) Jonathas DE MELLO (UNEP), Robert RODRIGUEZ (UNEP),Didier SALZMANN (UNEP)
Let us know how this tool can be useful to your own context, if there are areas of improvement or if you would like to chat about joint developments
