All-as-Vision: A Content-Native Exercise Interface for Transferable Knowledge Tracing

All-as-Vision represents exercises as visual content to support cross-domain knowledge tracing without task-specific feature engineering.

Recommended citation: Wang, Z., Qing, Q., Jiang, H., Hou, M., Luo, R., Zhang, X., & Xu, Z. (2027). "All-as-Vision: A Content-Native Exercise Interface for Transferable Knowledge Tracing." AAAI 2027. Conference submission.