Integrating Personalized Learning into Online Education through Content Aggregation, Data Mining, and Reinforcement Learning

Prihar, Ethan

Etd

Integrating Personalized Learning into Online Education through Content Aggregation, Data Mining, and Reinforcement Learning

Public Deposited

Personalized learning stems from the idea that students benefit from instructional material tailored to their needs. While the concept of giving each student the content that helps them learn the most is straightforward, implementing this at scale requires overcoming a gauntlet of challenges. One must aggregate a breadth of content such that enough variety exists to support each students’ specific preferences, calculate quantifiable aspects of students’ behaviors and traits that correlate with which content is most effective for them, design metrics that accurately measure learning, and create algorithms that can learn the relationships between students’ features and the effects of different content on their learning across thousands of students in real time. This dissertation discusses different approaches for collecting, interpreting, and recommending instructional content to students with a focus on learning interpretable insight that can inform educational pedagogy outside of online learning as well as within it. Ultimately, we designed a content recommendation algorithm that performed equivalently or better than similar existing algorithms while also allowing for unbiased statistical analysis of the data.

Creator