Laying a Foundation for the Graphical Course Map
Linda DuHadway, Thomas C. Henderson · 2016 · Proceedings of the 13th International Web for All Conference (W4A) · doi:10.1145/2899475.2899486
Summary
This paper from the University of Utah presents ENABLE, a system that transforms traditional linear, text-based learning management system (LMS) course presentations into interactive graphical course maps. The authors argue that current LMS platforms like Canvas impose two unnecessary constraints on online education: a text-based linear presentation of materials organized by chronology, and rigid temporal ordering that forces all students through content at the same pace. These constraints are particularly limiting for learners with diverse abilities who may respond better to visual representations or need flexibility in timing and sequencing. The ENABLE system extracts data from existing LMS courses via the Canvas API, identifying relations between learning items — temporal precedes (one item comes before another), includes (a unit contains an item), occurs in (a topic appears in an item), and prerequisite (educational value in doing one before another). These relations are used to build a course map represented as a graph where nodes are learning items and edges are relations. The key insight is that only prerequisite relations are truly constraining — temporal ordering can be relaxed in online settings, opening up many possible paths through the material while maintaining educational integrity.
Key findings
The system was tested using artificial student agents with configurable characteristics (intelligence, work ethic, background, distractibility) traversing course graphs in varied orders. Simulations revealed that agents using an equal-time strategy across all accessible items achieved mastery faster than those sequentially completing one item at a time before moving on — suggesting that the traditional linear organization of course material may actually slow learning. Predictive models (mixed linear, Bayesian network, and linear regression) trained on existing course data achieved 67-77% accuracy in predicting student scores. Critically, restricting these models to only prerequisite relations (rather than full temporal precedes relations) reduced accuracy by only 2-5%, demonstrating that data from existing linear courses can train useful predictive models for flexible-path delivery. A learning model based on mastery tracking with Kalman Filter estimation was developed to compute relative difficulty of learning items, providing instructors insight into workload balance. The interactive course map display allows nodes to be reorganized by topic, exam, item type, or prerequisite chains while maintaining relational integrity.
Relevance
This work addresses an often-overlooked accessibility barrier in online education: the rigid temporal and sequential structure inherited from traditional classrooms. While online learning is frequently promoted as accessible, most LMS platforms simply replicate classroom constraints digitally. For students with disabilities, chronic illness, or other circumstances that make synchronous, lockstep progression difficult, removing unnecessary temporal ordering could be transformative. The graphical course map approach also benefits visual learners and those who struggle with text-heavy, linear interfaces. For accessibility practitioners working in educational technology, the paper raises an important design question: which constraints in an LMS are pedagogically necessary (prerequisites) versus merely conventional (temporal ordering)? The finding that linear organization may slow learning provides evidence-based motivation for more flexible course delivery. The work remains at a foundational stage without user testing, which the authors identify as a key next step.
Tags: education accessibility · personalized learning · data visualization · learning management systems · machine learning · inclusive design