Technological Contribution of the Federal University of Uberlândia to the SDGs: a Patent Study

Abstract

This study aims to classify patents linked to the Federal University of Uberlândia (UFU) according to the Sustainable Development Goals (SDGs) of the 2030 Agenda, through the application of Artificial Intelligence (AI) techniques. The methodology integrates natural language processing and supervised learning approaches, using semantic embeddings generated by the BERT model. Initially, the patent texts (title and abstract) were vectorized and semantically compared to the SDG statements, enabling automatic classification based on similarity. In parallel, supervised classification models (Random Forest, Logistic Regression, and XGBoost) were trained to predict the SDGs from the embeddings, with evaluation through cross-validation and the use of precision, recall, and F1-score metrics. The results revealed the formation of partial thematic clusters, with a higher density of points around certain SDGs, such as SDG 1 (No Poverty), SDG 5 (Gender Equality), and SDG 12 (Responsible Consumption and Production), suggesting that these categories concentrate a greater volume of technological production with shared linguistic characteristics. This study was supported by FAPEMIG.