Predicting the extent of positive and negative contributions of companies to the UN Sustainable Development Goals with NLP

Abstract

With the environmental, social and governance framework becoming more and more integral to the progress of an organization, companies benefit from identifying how they score in ESG metrics and how much they contribute to the Sustainable Development Goals. However, reporting on and analysing this data effectively can be a complicated, time-consuming, and expensive task. In addition, ESG ratings alone do not always allow a company’s stakeholders to track the said business’s contribution to the United Nation Sustainable Development Goals as they do not directly map into the 17 Goals. (Amel-Zadeh et al., 2021) proposes a solution to align environmental, social and governance with Sustainable Development Goals already but it only measures the positive contribution (on a binary scale) and predicts the alignment and does not predict the extent of the contribution or identify negative contribution (on a continuous scale). This research proposes using Natural Language Processing and machine learning techniques to fill this gap by developing a novel approach that applies a continuous scale to measure the extent of a company’s contribution to the Sustainable Development Goals through the environmental, social and governance metrics presented in the organization’s sustainability-related documents. The proposed system will calculate ESG score, and calculate the contribution score for each Sustainable Development Goal and then classify the contribution as positive or negative.