Abstract

Sustainable Development Goals (SDGs) are a global commitment agreed upon by all Member States of the United Nations (UN) as the 2030 agenda, to realize a better and more sustainable world life. Indonesia, which is one of the UN member countries, must also commit to realizing the achievement of the 17 goals in the SDGs. Universities have an important role in achieving the SDGs goals, research aims to achieve this by identifying the track record of research and scientific publications that have been carried out by universities. GARUDA (Garba Rujukan Digital) is the official repository portal of the Indonesian government that provides access to various sources of scientific knowledge, including journals and proceedings published by universities in Indonesia. To find out the extent of the role of universities, it can be done by identifying the SDGs category of scientific publications conducted by researchers and indexed by GARUDA. This study aims to develop an SDGs category identification system using the DistilBERT model which is a distillation version of BERT. From the testing conducted on the testing data from GARUDA, the results of accuracy, precision, recall, and f1-score were obtained at 0.82. These results indicate that the model has quite good performance in identifying SDGs categories based on abstract data from GARUDA-indexed scientific publications written by university researchers in Indonesia.