Direct optimization of learning outcomes in the OptLearn algorithm and its scalability
There are several ways to design the OptLearn algorithm. An important property of a particular approach is whether or not it can directly optimize student learning outcomes. For example, if we track the student’s clickstream and assess their attention, we should be able to use this as an indicator of their engagement and feed it into the optimization.
Example:
The OptLearn algorithm proposed by Partner UL scales directly to newly estimated learning outcomes and allows us to define explicit criteria for when the concept is mastered, when questions should be repeated, etc.
Reference:
[1] A. Košir et. al. OptLEarn algorithm technical report, 2023.
[2] R. Ogawa and E. Collom, “Educational indicators: What are they? How can schools and school districts use them?,” Tech. Report ED432811, California Educational Research Cooperative, School of Education, University of California, Riverside, Nov 1998.
[3] J. Caspersen, J.-C. Smeby, and P. O. Aamodt, “Measuring learning outcomes”, European Journal of Education, vol. 52(1 ), pp. 20–30, Mar 2017.
[2] R. Ogawa and E. Collom, “Educational indicators: What are they? How can schools and school districts use them?,” Tech. Report ED432811, California Educational Research Cooperative, School of Education, University of California, Riverside, Nov 1998.
[3] J. Caspersen, J.-C. Smeby, and P. O. Aamodt, “Measuring learning outcomes”, European Journal of Education, vol. 52(1 ), pp. 20–30, Mar 2017.
Author of the tip:
Andrej Košir
University of Ljubljana