Degree Curriculum Contraction: A Vector Space Approach

This paper introduces a curriculum contraction technique in the context of university degree programs using a vector space embedding approach. We propose a way to model degrees and majors and define a contraction that takes the curriculum of a degree program and defines a smaller set of courses to approximate it. For example, a computer science degree curriculum could be generated that takes three years to complete instead of four (a 75% contraction). We use seven years of student enrollment data from a public university to train our embedding model. The most popular majors at the university, and their corresponding minors, are used to evaluate the validity of this contraction approach where minors are treated as major contractions.