<![CDATA[Mountain hiking activities in Indonesia have increased significantly; however, information on hiking routes is often dispersed informally and remains subjective, increasing the risk of a mismatch between hikers’ abilities and route characteristics. This study develops a hiking route recommendation system using a content-based filtering approach powered by Sentence-BERT (SBERT) to extract semantic features from narrative route descriptions, which are integrated with numerical GPX attributes. Similarity is computed using cosine similarity. Experimental evaluation on 39 hiking routes on Java Island shows an average Precision@5 of 0.60, outperforming TF-IDF (0.25) with a 140% improvement. The system proves effective in capturing implicit semantic relationships, thereby providing more relevant and context-aware recommendations.]]>
Copyrights © 2026
