TrueLearn

The code that allows you to run truelearn experiments that tests the performance of TrueLearn: A bayesian learning strategy that models lifelong learner engagement incorperating background knowledge of the learner and novelty of educational resources.

Citing the paper

This work is related to the paper TrueLearn: A Family of Bayesian Algorithms to Match Lifelong Learners to Open Educational Resources published at the Thirty-forth AAAI Conference on Artifical Intelligence, 2020 at New York, NY, USA.

The bibtex entry for the publication is as follows:

@inproceedings{truelearn2020,
	author = {Bulathwela, S. and Perez-Ortiz, M. and Yilmaz, E. and Shawe-Taylor, J.},
	title={TrueLearn: A Family of Bayesian Algorithms to Match Lifelong Learners to Open Educational Resources},
	booktitle = {AAAI Conference on Artificial Intelligence},
	year = {2020}
}

About the files

All the useful files in this module incorporate different algorithms used in the experiments

• naive_baseline_models.py: file contains the programming logic of persistence, majority models
• multi_skill_kt_models.py: file contains the programming logic of multi-skill knowledge tracing models
• fixed_depth_truelearn_models.py: file contains the programming logic of fixed-depth truelearn models
• truelearn_models.py: file contains the programming logic of TrueLearn Novel model

Additionally, there are two files that are used to run the experiment:

• run_sequencial_trueskill_baseline.py: file that is used to run Vanilla TrueSkill baseline and TrueLearn dynamic-depth as they are sequential models
• run_experiments.py: file that is used to run all the other experiments

Other helper files:

• utils.py: file contains required utility functions
• analyse_results.py: file used to produce precision, recall, accuracy and F1 metrics from the algorithms