Learn Structure Underlying Problem-Solving in Math

ASTRA (AI-based STRategy Discovery) is a tool analyzing math-learning strategies in big-data.

When we imagine how students learn problem-solving skills, it is apparent that this is dependent on a combination of complex factors such as opportunities to work on similar problems, general expertise in the topic, other cognitive abilities, social factors, etc. While researchers have tried to understand this process over a long period, revolutionary advances in AI offers a unique opportunity to analyze strategies in problem-solving at scale and thus make progress towards personalized education.

Our group is developing ASTRA (AI-based STRategy Discovery) where the goal is to learn Large Strategy Models (LSMs) that encode Math problem solving strategies. Analogous to Large Language Models (LLMs) that understand language structure, our goal is learn structure underlying problem solving in Math. Using this, we hope to build a general research platform that can be utilized to solve several challenging tasks in education such as inferring mastery, designing personalized interventions, etc.

Through generous support from the Bill and Melinda Gates Foundation as well as the Learning Agency, we will develop novel AI models in ASTRA that encode strategies using large-scale datasets collected from partner organizations and provide open access to these models. Also, we will also develop studies which will demonstrate the potential of ASTRA in real-world environments through our partner organization Carnegie Learning, the developer of MATHia, a leading adaptive instructional system for Math learning.

Dr. Deepak Venugopal, associate professor of Computer Science whose expertise is in AI and their applications, is the PI on this project. The Co-PI, Dr. Vasile Rus, professor of Computer Science, is a leading expert in AI-based educational technologies. The team from Carnegie Learning consists of Dr. Steve Ritter, co-founder and chief scientist for Carnegie Learning, Dr. Steve Fancsali, the director of advanced analytics at Carnegie Learning and Dr. April Murphy who is leading the MATHia testing platform.

For more information on this project, contact Venugopal at dvngopal@memphis.edu.