Pengembangan Sistem Evaluasi Adaptif Berbasis LLM dalam Smart Tutor Menggunakan Bloom’s Taxonomy

Nicholas Rudy (Program Studi Teknik Informatika, Fakultas Teknologi dan Rekayasa Cerdas, Universitas Kristen Maranatha)
Hapnes Toba (Program Studi Teknik Informatika, Fakultas Teknologi dan Rekayasa Cerdas, Universitas Kristen Maranatha)

Journal of Smart Technology and Engineering
Vol. 2 No. 2 (2026):

This research aims to improve adaptive learning systems by developing a Smart Tutor platform enhanced with Large Language Models (LLMs). Traditional systems that rely on keyword matching or cosine similarity often fail to accurately assess student responses due to their lack of semantic understanding. To address this, we propose a semantic-based evaluation system using BERTScore and TF-IDF to analyze student answers more contextually.