Coursera has emerged as a formidable competitor in the online education space, racking up 24 million registered learners with access to 1,900 courses. But with that growth in volume comes new challenges. With so much content, it’s difficult for users to find the videos they need. Today, Coursera is formally announcing a new search tool enabling users to search directly for skills they want to learn and receive recommended content from the company.
Powered by machine learning, the new skill search was built off a skills graph that maps skills to content. A classifier then predicts the likelihood that a respective skill will be taught in a given course. The images below depict before and after shots — the new search feature surfaces courses that involve “P-Values,” a skill that is so granular that it likely isn’t mentioned in the descriptions of courses that touch on it.
[gallery columns="2" ids="1453740,1453739"]
Tom Willerer, chief product officer at Coursera, was previously director of product innovation at Netflix, the company that made content recommendation sexy. At a high level, skill search works in a similar way to Netflix’s framework for TV and movie recommendations.
Machine learning made Netflix possible; drawing users in with costly popular shows and keeping them hooked with less expensive, often overlooked, shows was a core thesis for the company — at least until it started creating its own content en masse.
Though Coursera doesn’t need machine learning in the same way, the technology does offer the potential to help the company measurably differentiate from competitors like Udacity and Khan Academy. The more preemptive leg work Coursera can do for its learners, the more likely they are to find the right content on their first attempt and stick around with high levels of engagement.
Coursera is just starting to dip its toes into machine learning. To make the biggest impact on the user experience at scale, the startup will need to get a lot more granular. The team behind today’s release says that’s exactly what they want to do for immediate next steps, ultimately offering search at the level of an individual video rather than an entire course.
We are a ways off, but it’s not hard to imagine a world where Coursera could leverage its enterprise business to map learners directly from skill to job. As of now, Coursera’s relationship with enterprises is focused on developing better content and collecting analytics from on-the-job learning.
A natural extension could be collaboration with enterprises to identify the skills necessary for given roles. Making online learning more active and personalized, the company could directly mentor students to help them accomplish tailored goals.