DICE, a Parse-Tree Based Automatic Programming Assessment System with Test Driven Development

• Main Authors: Li-Ren Chien, 簡立仁
• Other Authors: DANIEL J. BUEHRER
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/35333336139497337566

博士 === 國立中正大學 === 資訊工程所 === 97 === JUSTICE DICE, the personification of justice, a goddess of Roman mythology, was taken as the name of our Automatic Programming Assessment System (APAS) since 2006. We aimed to build an on-line auto-grading system for judging programming assignments. Justice is one of the most important characteristics in DICE, which has been used in blended learning in the classroom under an instructor’s supervision. We emphasize the automatic assessment system and plagiarism detection mechanism to guard against injustice by punishing injustice and rewarding virtue. BOUNDARY Confucius argued that “Education knows no boundaries.” As a specialized learning management system, DICE plays the role of delivering general purpose assignments and exercises. DICE checks programming exercises and conceptual knowledge in a specific topic area. DICE is designed as a stand-alone service to support the education across space and time. Practice Confucius claimed that “To learn and then have occasion to practice what you have learned—is this not satisfying?” DICE is treated as an augmented training approach to support the classical lecture-based teaching for computing. The result is the students put more effort on improving their programming skills via practice in class. Sequence Confucius argued for “Improvement in proper sequence.” We have made use of the method of ‘Test Driven Development” (TDD) which was introduced in software engineering. The DICE TDD classifies the training material from exploration to instruction into 16 levels. It supports DICE to help check student’s code, in terms of both his programming skill and his conceptual knowledge of a specific topic. Adaptation Confucius said, “Teach students in accordance with their aptitude.” Our results indicate that the influence of Kolb learning style has different significance on learning outcomes in TDD verse non-TDD. An adaptive learning model was proposed to adapt to students with different TDD training materials based on their personality differences. We also developed a Typed Mind Map as a common data model and knowledge representation in DICE. DICE promotes learning performance up to 11% better than a non-DICE approach, while a DICE with TDD model improves the learning performance to 51% more than the pure DICE approach.