DS503_OPTION-KNOWLEDGE CONCEPT AWARE KNOWLEDGE TRACING

Dataset

• Ednet
• Eedi_a
• Eedi_b

Download Link

• dataset for model train
• dataset for preprocessing

---

Model

• dkt: Deep Knowledge Tracing
• dkt+: Addressing Two Problems in Deep Knowledge Tracing via Prediction-Consistent Regularization
• kqn: Knowledge Query Network for Knowledge Tracing
• sakt: A Self-Attentive model for Knowledge Tracing
• saint: Towards an Appropriate Query, Key, and Value Computation for Knowledge Tracing
• atkt: Enhancing Knowledge Tracing via Adversarial Training

---

Project Structure

├── ckpts
├── data
│   ├── ednet
│   |   ├── method_1
│   |   └── method_2
│   ├── eedi_a
│   |   ├── method_1
│   |   └── method_2
│   └── eedi_b
│       ├── method_1
│       └── method_2
├── preprocessing
│   ├── data
│   │   ├── ednet
│   |   |   ├── method_1
│   |   |   └── method_2
│   │   ├── eedi_a
│   |   |   ├── method_1
│   |   |   └── method_2
│   │   └── eedi_b
│   |       ├── method_1
│   |       └── method_2
│   ├── method1_preprocessing(py)
│   └── method2_preprocessing(ipynb)
│   
├── train.py
├── dkt.py
├── dkt_plus.py
├── kqn.py
├── sakt.py
├── saint.py
├── atkt.py
└── utils.py

Folder	Usage
ckpts	save checkpoint model
data	data for model training
preprocessing	data, code for preprocessing

---

Environment

    conda create -m kt
    conda activate kt
    conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia
    pip install -r requirements.txt

---

User Guide

Preprocessing (Optional)

1. Download dataset for preprocessing

2. Execute preprocessing code in preprocessing folder

   Method 1

   • ednet

   python method1_preprocessing_ednet.py
   

   • eedi_a

   python method1_preprocessing_eedi_a.py
   

   • eedi_b

   python method1_preprocessing_eedi_b.py
   

   Method 2

   Execute method2_preprocessing ipynb file for each dataset in preprocessing folder

Train

1. Download dataset for model train

2. Modify config.json

3. Execute train.py

   python train.py --model_name dkt --dataset_name eedi_a --method_name method_1 --option no
   

   option

   • model_name
     • dkt, dkt+, kqn, sakt, saint, atkt
   • dataset_name
     • eedi_a, eedi_b, ednet
   • method_name
     • method_1, method_2
   • option
     • no, no_kc, no_option, yes