Tags: Curriculum maps
Interactive visualizations of learning pathways over the MIT undergraduate curriculum.
From left to right: Subject units distributions; Prerequisite relationships; OpenCourseWare coverage.
Network maps created using the Rhumbl Mapping Engine →
Every educational institution has curriculum data — typically represented in some kind of list or table form. The typical entity in a curriculum is a course, which has information associated with it such as credits, prerequisite requirements, a description, etc. Current practices are to store this information as a table, where relationships are flattened to attributes in columns. This flattened format makes it difficult to do the analytics and visualization that supports educational decision-making and innovation.
Our curriculum mapping model formalizes the process of modeling curricular data in a structured model that explicitly models the connections among curricular entities.
We first define the different types of entities: Course, Department, Institution. Each entity of these various types is modelled as a vertex. They are assigned attributes, e.g. name, course number, schedule information, website listing, etc.
Next, we define different types of relationships among the entities. A basic curriculum model has three types of relationships: has-parent-of, has-prerequisite-of and has-corequisite-of. The has-parent-of relationship is a directed relationship that specifies organizational hierarchy in the curriculum. The has-prerequisite-of and has-corequisite-of relationships are directed; they exist between courses to define prerequisite and corequisite relationships. These relationships are modelled as directed edges between the appropriate vertices in the graph model.
Model ontology of the curriculum graph model
This way of modeling provides a powerful basis for visualization and analytics. For example, we analyze the curriculum graph structure to ask questions such as "For how many downstream classes is this class a prerequisite?" or "What are flexible learning pathways that lead to this class?"
Or we can join other data sets to conduct analytics that cut across the curriculum: "What are the hotspots of dropouts across the curriculum and how does that affect downstream classes?" "What would be the institutional-wide impact of a schedule shift in a gateway class?" etc.
