AI Based Mind Map Programming Tutor

The increasing adoption of AI-driven programming assistants has significantly transformed the landscape of coding education. Despite these advancements, most existing intelligent tutoring systems emphasize answer generation rather than diagnosing learners’ underlying misconceptions and providing adaptive remediation. This paper introduces the Mind-Map Programming Tutor (MMPT), a novel system that employs cognitive graphs to model learners’ evolving knowledge states. In contrast to traditional AI tutors that require pre-annotated student data for training, MMPT leverages zero-shot reasoning to infer misconceptions directly from student code submissions. It dynamically constructs a cognitive error graph that captures conceptual misunderstandings, knowledge gaps, and logical inconsistencies.