A Multi-Ontology Approach for Deforestation Issue Analysis in Indonesian Digital Media

Abstract

The increasing volume of online news has made the identification of deforestation issues more challenging, particularly because conventional text classification methods rely mainly on statistical features and often fail to capture semantic relationships between terms. This study proposes a semantic text classification approach for Indonesian deforestation news by integrating DBpedia and a lexical ontology to enrich document representation before classification using the Support Vector Machine (SVM) algorithm. News articles were first preprocessed through case folding, tokenization, stop-word removal, and stemming. Semantic features were then generated through multi-ontology mapping and transformed into feature vectors for SVM classification. The proposed approach was evaluated using 91 testing documents with Accuracy, Precision, Recall, and F1-score as performance metrics. The results achieved an accuracy of 87.9%, with 88.5% precision, 87.8% recall, and an F1-score of 88.0%. These findings indicate that ontology-based semantic feature extraction improves document representation by capturing conceptual and lexical relationships that cannot be represented by statistical features alone. The proposed approach provides an effective and interpretable solution for classifying Indonesian deforestation news into four issue categories.