Neuro-Symbolic Models: A Scalable, Explainable Framework for Strategy Discovery from Big Edu-Data
Predicting student problem-solving strategies is a complex problem but one that can significantly impact automated instruction systems since they can adapt or personalize the system to suit the learner. While for small datasets, learning experts may be able to manually analyze data to infer student strategies, for large datasets, this approach is infeasible. While Deep Neural Network (DNN) based methods such as LSTMs can be applied for this task, they have drawbacks such as long convergence times for big datasets, and like DNN-based methods in general, have the inherent problem of overfitting the data. To address these issues, we propose a general Neuro-symbolic framework for strategy prediction, where we combine the strengths of symbolic AI (that can encode domain knowledge) with DNNs. We outline several possible benefits of this framework and demonstrate its potential in scalable learning from large educational datasets.
CEUR Workshop Proceedings
Venugopal, D., Rus, V., & Shakya, A. (2021). Neuro-Symbolic Models: A Scalable, Explainable Framework for Strategy Discovery from Big Edu-Data. CEUR Workshop Proceedings, 3051 Retrieved from https://digitalcommons.memphis.edu/facpubs/3014