Online learning startup Coursera today debuted CourseMatch, an AI tool that matches classes in schools’ on-campus course catalogs to relevant offerings in Coursera’s catalog. It’s a part of the startup’s Coronavirus Response Initiative that launched in early March, and Coursera says it’s intended to enable universities to facilitate learning as governments mandate the shutdown of non-essential institutions in response to COVID-19.
“As universities go live using our offering, they urgently need an easy solution to help identify courses on Coursera that most closely match each course in their on-campus catalogue,” Coursera VP of data science Emily Glassberg Sands and data scientist Marianne Sorba wrote. “This is a hard problem to solve, especially across thousands of universities and millions of on-campus courses. Two weeks back, the Data Science team at Coursera started to work on developing a solution — CourseMatch — that enables faculty to go online quickly, without much human curation.”
More than 91,000 U.S. public and private schools were closed as of March 25, affecting more than 50 million students. And many of these schools have been beset by challenges that Coursera believes CourseMatch can solve, like how to provide instruction to students at different learning levels.
CourseMatch ingests courses in English or any of the over 50 languages (both translated and subtitled) available on Coursera. It then uses natural language techniques and pretrained word embeddings (words from a vocabulary mapped to vectors of real numbers) to find Coursera courses most semantically similar to each course in an on-campus catalog. For on-campus courses, the algorithms consider the course title and description from the publicly available catalog alongside more detailed syllabi and learning objectives (where provided). As for courses on Coursera, the algorithms consider the full-text corpus, including the course title and description as well as the lecture transcripts, assignments, and assessments.
Finally, CourseMatch returns up to five of the most relevant Coursera courses for each on-campus course along with corresponding relevance scores, which are normalized within colleges and universities; higher scores represent stronger matches. Coursera says that already, CourseMatch has matched more than 2.6 million curricula across 1,800 schools including Johnson C. Smith University and Brown University in the U.S.; IESE in Spain; and ACITE in India.
“Our goal is to help every college and university in the world move online as quickly and easily as possible,” Glassberg Sands and Sorba wrote. “This is our first iteration of CourseMatch, and we look forward to refining our machine learning models based on … feedback. We invite institutions to try this solution as [they] quickly scale programs online during the COVID-19 crisis.”
CourseMatch builds on Coursera’s existing Coursera for Campus tool, which is designed to help universities — even those that aren’t partners — supplement their course offerings with Coursera’s content. Colleges enrolled in the Coursera for Campus program gain access to more than 3,600 programs that they can integrate into student, faculty, and staff curricula. Preconfigured collections are available in domain areas like engineering, business, data science, law, health, arts, and design, all of which universities can offer directly or swap out with others from Coursera’s broader catalog (or with courses they’ve authored themselves).
Coursera for Campus customers can also tap Coursera’s online assessment and content creation tools, including analytics dashboards like Course Progress and Skills Index and the recently launched Coursera Labs for hands-on assignments. Faculty can publish private lessons, hands-on projects, assessments, and courses for students and alumni alike.
Coursera says that since March 12, over 3,000 colleges and universities have activated Coursera for Campus programs.
The audio problem: Learn how new cloud-based API solutions are solving imperfect, frustrating audio in video conferences. Access here