/cl4kt

This is our implementation for the paper "Contrastive Learning for Knowledge Tracing" (TheWebConf 2022).

Primary LanguagePython

Contrastive Learning for Knowledge Tracing

This is our implementation for the paper Contrastive Learning for Knowledge Tracing (TheWebConf 2022).

To run CL4KT, please prepare the configuration file (configs/example.yaml) and the raw dataset (e.g., datatset/algebra05/data.txt, datatset/assistments09/data.csv, etc.).

For example, the algebra05 dataset comes from the KDD Cup 2010 EDM Challenge. Datasets need to be downloaded and put inside each corresponding data folder in dataset.

Please use the following script to run data preprocessing:

python preprocess_data.py --data_name algebra05 --min_user_inter_num 5

Please use the following script to run the CL4KT model:

CUDA_VISIBLE_DEVICES=0 python main.py --model_name cl4kt --data_name algebra05 --mask_prob 0.5 --crop_prob 0.3 --permute_prob 0.5 --replace_prob 0.5 --reg_cl 0.1