Mapping Curricula to Skills and Occupations Using Course Descriptions

Aligning STEM curricula with skills and occupations benefits students, employers, and educational institutions. However, manual curriculum mapping to assess alignment is labor-intensive and does not scale well. This study evaluated the use of natural language processing (NLP) to automate mapping curricula to skills and occupations. An open-source NLP tool was applied to extract skills from course descriptions in data science programs at two institutions, and from occupation descriptions from the US Department of Labor's O*NET database. Occupations were ranked by similarity between curriculum and occupation skill sets. Occupations were classified as relevant or nonrelevant to data science and analytics, and the standard information retrieval metric of precision@N was calculated. Skill extraction identified an average of 5.3-10.4 skills/course, and precision@5 was 60%-100%. There appears to be strong potential for using NLP of course descriptions to map curricula to skills and occupations, at least as a starting point for manual refinement.