An automated assessment system for data structures and algorithms with diverse question types and knowledge-driven evaluation

Tóm tắt

This paper proposes an automated assessment system for Data Structures and Algorithms (DSA) that supports eight diverse quesƟon types-MulƟple-Choice QuesƟons (MCQ), Fill-in-the-Blank Number (FN), Fillin-the-Blank Number Sequence (FNS), Fill-in-the-Blank Programming Syntax (FPS), Fill-in-the-Blank Short String (FSS), Fill-in-the-Blank Expression/Formula (FE), Matching Pairs (MP), and Programming Exercises (PE)-across seven difficulty levels (Easy to Expert). Leveraging Bloom's Taxonomy and psychometric parameters (difficulty: p, discriminaƟon: d), the system uses knowledge matrices to integrate Course Learning Outcomes (CLOs), six content chapters, and cogniƟve levels, enabling adapƟve test generaƟon for quizzes, midterms, and exams. Automated scoring employs exact matching for MCQ/MP/FN/FNS, paƩern matching for FPS/FSS, symbolic evaluaƟon for FE, and test-case evaluaƟon for PE. The system tracks knowledge progression across tests, per topic, and per CLO, providing comprehensive feedback. IniƟal implementaƟons have shown the system's ability to improve assessment quality, decrease grading efforts, and facilitate data-driven teaching approaches in technical fields such as DSA.