Penerapan Model Support Vector Machine Pada Kasus Klasifikasi Teks Berdasarkan Tujuan SDGS Ke Tiga, Empat, Dan Enam

Abstract

Text classification is a branch of Natural Language Processing (NLP) that enables computers to understand, interpret, and respond to text in a comprehensible language. Classifying texts based on the Sustainable Development Goals (SDGs) is crucial because monitoring the progress of SDGs remains a challenge. Previous studies have shown that text classification techniques using the BERT model have proven effective in classifying texts based on SDG goals. This research utilizes data sourced from the OSDG community website. The method employed is the Support Vector Machine Multiclass (SVM) model and TF-IDF word representation. This research aims to classify texts based on the Sustainable Development Goals (SDGs), specifically focusing on goals three, four, and six., evaluate the model's performance based on the F1-Score metric, and determine the optimal values for the hyperparameters regularized constant  and gamma  in the RBF kernel. The results of this research yielded a default F1-Score of 97.95% and a post-tuning F1-Score of 97.95%, with the optimal values of C=1, gamma=1, and kernel=rbf.