PENGEMBANGAN SISTEM REKOMENDASI BERBASIS KNOWLEDGE GRAPH PADA APLIKASI WEB PERSONALIZED INTELLIGENT TUTORING SYSTEM UNTUK PEMBELAJARAN PEMROGRAMAN

• Main Author: Dwi Fadhilah, Rania
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/82410
• Institution: Institut Teknologi Bandung

In facing the shortage of digital talent in Indonesia, innovations are required to accelerate the fulfillment of increasing needs. CodeBuddy.ai is an intelligent tutoring system (ITS) web application that serves as a platform for learning basic C++ programming for beginners. This ITS is equipped with personalization in the form of learning paths tailored to students' abilities through the use of a knowledge graph (KG)-based recommendation system. The recommendation system based on knowledge graphs is an approach used by this ITS because it can reflect semantic representation and adapt to the needs of the ITS. This study focuses on comparing two models within the KG-based recommendation system: semantic similarity calculation that uses semantic weighting for the calculation of entity similarity paths and random walk with KG embedding that explores relationships more broadly through its embedding results. Both models are evaluated for their ability to identify and recommend learning paths relevant to the programming learning domain in ITS. Experimental results show that the random walk with KG embedding has advantages in detecting distant entity relationships more effectively, making it superior in formulating accurate recommendations compared to the semantic similarity calculation model.