Provide line-level KC-based rationales to make recommendations auditable and pattern-focused¶

Strategy · All strategies
Evidence · no claims cited

Description¶

Because similarity is computed over pattern-based KCs, the system decomposes the cosine score into per-KC contributions and traces high-contributing KCs back through the pipeline to high-attention subtree instances and their code spans, highlighting them "with color intensity proportional to the underlying attention weight". For students, highlighted regions direct attention to abstract decision-making schemas to support transfer; for instructors, contribution breakdowns provide an auditable account of matches, help diagnose failure modes, and support maintenance of the KC inventory.

Design Implications¶

Context¶

Requirements¶

• Explanation quality depends on the attention weights assigned by the framework and the KC clustering resolution

Constraints¶

• Highly overlapping KCs can produce adjacent or redundant highlights, as the authors state; the pipeline's topic-label training step limits direct application to unlabeled corpora

Target Learners¶

• introductory programming students
• instructors curating programming content

Target Learning Goals¶

• internalizing general structures of conditional reasoning and other programming patterns
• auditing and refining content repositories

Affordances¶

• Pattern Based Kc Representation Framework

Related Strategies¶

Examples¶

-

Key Sources¶

• Muntasir Hoq, Griffin Pitts, Zhangqi Duan, Arun Balajiee Lekshmi Narayanan, Mohammad Hassany, Andrew Lan, Peter Brusilovsky and Bita Akram. (2026). Automated Recommendation of Programming Learning Content Using Pattern-based Knowledge Components. CSEDM'26: 10th Educational Data Mining in Computer Science Education Workshop. https://arxiv.org/abs/2607.05409