In Quest for Meaning: Towards a Common Understanding of the 2030 Agenda?

Recent developments in Natural Language Processing (NLP) are revolutionizing knowledge management (Hu et al. 2023). The generation of Large Language Models, like ChatGPT, breaks down barriers between different types of languages (Naveed et al. 2023). In a few seconds, it is now possible to edit complex programming codes from a prompt written in vernacular language (Xu et al. 2022).

As a specific branch of NLP, classification involves the recognition and mapping of references to a specific topic within a text. This technique is useful for analyzing large corpuses of documents. Conceptually, the classification of Sustainable Development Goals (SDGs) is a particularly technical case. Adopted in 2015, the 2030 Agenda constitutes a common framework for approaching and implementing human development policies while respecting environmental boundaries. The 2030 Agenda is structured around 17 objectives, which are themselves broken down into 169 targets. In this regard, training an NLP model for SDG classification requires a detailed understanding of the specificities of each objective, as well as their interactions.