New Framework Measures Computer Science Curriculum Alignment with Guidelines

Computer science undergraduate programs are guided by international curricular standards, which are updated periodically. However, institutions often lack a consistent and reproducible method to quantify how thoroughly their programs cover these guidelines and how that coverage changes over time. This research addresses this gap by proposing a human-in-the-loop pipeline for measuring curriculum alignment. The pipeline represents both the academic program and the guidelines (e.g., CS2013 and CS2023) as structured corpora. It then uses semantic retrieval to generate candidate matches between courses and knowledge units, which are subsequently confirmed by human judgment based on explicit coverage definitions. The study found that an ensemble retriever performed best, and surprisingly, a large language model underperformed a smaller sentence model for this task, emphasizing the importance of retriever selection. Applying this framework longitudinally to a BSc in Computer Science program, the research revealed that the program covered approximately 50% of knowledge units in both CS2013 and CS2023, indicating consistent structural gaps in areas like parallel computing. While competency articulation remained high, the program delivered recommended cognitive depth for 76% of units under CS2023, a decrease from 95% under CS2013, reflecting the newer guideline's elevated expectations rather than a program deficiency. This instrument provides a reusable method for educators to identify and address curriculum gaps.