<![CDATA[Assessment is an essential part of the learning process. Standardized assessment often fails to identify students’ cognitive abilities, including critical thinking skills, on an individual basis. Recent research shows growing potential in adaptive assessment through machine learning, particularly user-based collaborative filtering, though its application to critical-thinking instruments remains limited. This study analyzes research trends in the application of an adaptive assessment system based on User-Based Collaborative Filtering (UBCF) as a context-specific question recommendation mechanism that maintains measurement equity. The method used was a Systematic Literature Review following PRISMA 2020, searching four databases (Google Scholar, Semantic Scholar, IEEE Xplore, and GARUDA), yielding 25 final articles from 2,573 records. Findings reveal three results. First, UBCF is reliable as an educational recommendation mechanism, but its use for question selection in assessment remains unexplored, though Elo-rating and hybrid CF-IRT combinations have begun touching this domain. Second, adaptive assessment is consistently more effective than conventional assessment in measuring cognitive ability, provided the mechanism maintains conditional fairness. Third, critical thinking measurement in vocational programming education still relies on uniform conventional instruments, without integrating recommendation systems. This study concludes that UBCF can fill this gap through context-based question adjustment using similarity of student response patterns, while maintaining cognitive equity. Further research is needed for empirical validation in vocational classrooms. This study contributes a research gap map and a context-based adaptivity framework for UBCF-based assessment systems.]]>
