Implementation of Singular Value Decomposition with Constraint Base Approach for Internship Recommendation System for Vocational High School Students

Abstract

Vocational education in Indonesia, especially through Vocational High Schools, plays a crucial role in preparing students for the workforce. However, mismatches between student competencies and industry requirements often result in ineffective internship placements. This study focuses on SMK Negeri 2 Banda Aceh, where the internship placement process has been carried out manually and lacks an objective and personalized system. To address this challenge, a hybrid recommendation system was developed by combining Singular Value Decomposition with a constraint-based approach. The SVD method predicts student-industry compatibility by uncovering latent patterns in the rating data, while constraint-based filtering ensures that recommendations meet specific criteria such as major compatibility, skill alignment, and availability of industry capacity. The system was implemented as a web-based application using Python and MySQL, providing real-time recommendations with response times between one and three seconds. Testing with data from 344 students and more than 120 industry partners at SMK Negeri 2 Banda Aceh demonstrated the system’s ability to generate accurate and relevant recommendations. For example, although an industry with a predicted rating of 0.58 matched the student’s major and skills, it was not recommended due to full capacity. Instead, another industry with a lower predicted rating of 0.44 was recommended because it met all the required constraints. This system helps schools carry out internship placements more objectively, efficiently, and in alignment with student profiles and industry.