A novel Z-number based multi-stage assessment framework for problem-based learning in practical courses

Area of Science:

• Educational Technology
• Decision Sciences
• Artificial Intelligence in Education

Background:

• Standardized evaluations often conflict with personalized competency development in practical education.
• Problem-Based Learning (PBL) environments require nuanced assessment methods.
• Subjective and imprecise expert judgments pose challenges in educational evaluation.

Purpose of the Study:

• To propose a novel data-driven framework for enhancing curriculum assessment in PBL.
• To integrate Z-number theory and Multi-Criteria Group Decision Making (MCGDM) for uncertainty handling.
• To improve the robustness and reliability of educational assessment processes.

Main Methods:

• Developed a framework integrating Z-number theory with the Multi-Attributive Border Approximation Area Comparison (MABAC) method.
• Implemented a multi-stage assessment process: Pass check, Score determination, and Grading phases.
• Introduced a hybrid Entropy-CRITIC (Criteria Importance Through Intercriteria Correlation) weighting scheme using Z-numbers.

Main Results:

• Applied the framework to a case study with 24 students evaluated by peers, instructors, and industry experts.
• Sensitivity and comparative analyses confirmed the framework's robustness and reliability.
• The proposed method demonstrated a more transparent and structured evaluation procedure.

Conclusions:

• The Z-number enhanced MCGDM framework offers a reliable approach to PBL curriculum assessment.
• The method effectively addresses uncertainty and imprecision in expert evaluations.
• Further validation is needed for broader generalizability across different educational settings.