An Evidence-Based Fusion Framework for Program-Level Curriculum Evaluation in Computer Science: Standards Alignment and Cognitive Demand Across Ten Introductory Offerings

Program-level curriculum evaluation is central to quality assurance in higher education, yet remains labor-intensive when disciplinary standards must be interpreted across heterogeneous, publicly available curriculum documents. This is particularly evident in computer science, where the CS2023 framework defines 17 knowledge areas (KAs) but offers no operational instruments for examining how strongly each area is represented in actual offerings, or at what level of cognitive demand. Drawing on Bloom’s revised taxonomy and the principle of constructive alignment, and following educational design research, this study develops and evaluates an evidence-based fusion framework combining three complementary indicators: weighted lexical matching against the CS2023 KA structure, transformer-based semantic alignment using Sentence-BERT, and a four-band cognitive-demand classifier. The framework was applied to publicly available course materials (lecture transcripts, open textbooks, and course documents) from ten introductory computer science offerings spanning elite universities and open-access platforms, and validated through a structured expert review in which five domain specialists completed a 20-item, four-point instrument. Findings indicate that the framework surfaces comparable cross-offering profiles of disciplinary alignment and cognitive demand: foundational areas were consistently emphasized, while Security, Artificial Intelligence, and Society, Ethics, and the Profession ranked ninth or lower of the 17 areas in every sampled offering; cognitive-demand profiles could diverge even where alignment profiles converged. Experts’ ratings were positive overall (pooled item mean 3.4 of 4.0; the recommendation-to-proceed item was rated 4.0 by all five experts), while identifying validation depth as the principal refinement priority. The study contributes to curriculum analytics by integrating disciplinary alignment with cognitive demand in a single, interpretable instrument that complements rather than replaces expert academic judgment in program-level review.