Aligning national policies with the UN Sustainable Development Goals requires the analysis of large amounts of strategic, administrative and policy text. This process can be time-consuming when done manually, especially when documents need to be compared with many goals and sub-goals. AI and natural language processing can support this work by helping institutions identify connections between policy documents and sustainability priorities in a faster and more systematic way.

Within the project “The Use of AI to Automate the Rapid Integrated Assessment Mechanism and to Nationalize Sustainable Development Goals in Serbia,” an algorithmic solution was developed for measuring the degree of alignment between relevant national plans and strategies and the goals and sub-goals of the UN Sustainable Development Agenda. The solution was designed for the Serbian language, with relevance for other smaller and morphologically complex related languages.

The approach is based on a distributional semantic model that compares the meaning of each sustainable development goal with sentences and paragraphs from national plans and strategies. The AutoRIA model was trained on the textual content from national plans and strategies and evaluated on the first five Sustainable Development Goals from the UN Agenda. The results were comparable to those achieved for the English language, demonstrating the potential of AI and NLP methods to support policy analysis and strategic planning in Serbian and other smaller, morphologically complex languages.

The project was financed by the United Nations Development Programme (UNDP).