Improving education through automation.

Computer Research Institute of Montreal (CRIM)

Canadian education has advanced significantly beyond lectures delivered before a chalkboard. Technology has changed the educational landscape with videos, tutorials, and learning games, creating numerous new ways to train our next generation of doctors, engineers, teachers, and historians. These technologies afford reach to many more people in many more places, all the while reducing the overall cost of education.

Clearly, computer-assisted education has many advantages. But without the presence of a human instructor to ensure that students understand the content behind the video or online tutorial, how do we know if the instruction is effective?

The efficiency of technology-based education can be measured by examining videos of students interacting with a system. For most studies, this results in thousands of hours of video that needs to be transcribed and analyzed for student interactions – a process that can take ten times longer than the length of the video captured.

Automated video analysis

Researchers and experts at the Computer Research Institute of Montreal (CRIM), together with the Learning Environments Across Disciplines research group (LEADS), have created a host of valuable tools for educational researchers that allows them to assess the quality of the educational experience and to fine-tune educational technology for different learning styles. These tools create a consistent way to measure and improve computer-based education, and form the Research Software Platform known as VESTA (Video Evaluation System for Task Analysis).

VESTA distinguishes when the student is reading the screen or thinking aloud (which they are encouraged to do), and when they are paying attention or are distracted. These cues, along with the text transcription of the student’s interaction and automatic annotation of notable events (video transitions, facial changes, new individuals entering the scene) create catalogued and searchable results for each student video, making research on educational technology significantly easier.

Developing better doctors

Although VESTA is applicable to many types of research requiring video analysis, two current examples are in the medical field. One project helps train prospective doctors in assessing urgent care cases and the other helps them become better at delivering bad news. In both cases, student doctors are filmed as they work through a diagnosis or interact with patients. The resulting analysis not only helps the students become more efficient and empathetic doctors but also helps identify where the training material itself may be improved.

Contributions to other researchers

In addition to the VESTA Platform itself, a number of Software Services used within it have been made available to other researchers via the CANARIE Software Registry. These include tools for annotation storage, facial analysis, video transition detection, load balancing, multimedia file storage, speech segmentation, speaker discrimination, speech-to-text, and text-to-audio matching.

These new Services are available for reuse by researchers across many scientific disciplines, in the same way that VESTA made use of Services provided by other researchers through the CANARIE Software Registry.

Funding for the development of VESTA was provided through CANARIE’s Research Software Program.