# RCD

**Repository Path**: abc-pedicle/RCD

## Basic Information

- **Project Name**: RCD
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2024-01-05
- **Last Updated**: 2024-01-05

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# RCD: Relation Map Driven Cognitive Diagnosis for Intelligent Education Systems

This is our implementation for the paper of **RCD: Relation Map Driven Cognitive Diagnosis for Intelligent Education Systems** published on `SIGIR'2021`. [[Paper](https://dl.acm.org/doi/abs/10.1145/3404835.3462932)][[Presentation Video](https://dl.acm.org/action/downloadSupplement?doi=10.1145%2F3404835.3462932&file=RCD.mp4)]

Author: [Weibo Gao](https://scholar.google.com/citations?user=k19RS74AAAAJ&hl=zh-CN), [Qi Liu](http://staff.ustc.edu.cn/~qiliuql) et al.

Email: weibogao@mail.ustc.edu.cn

## Environment Settings
We use Torch and DGL as the backend. 
- Torch version:  '1.7.1'
- DGL version: '0.6.1'

## Example to run the codes
The instruction of commands and take Junyi dataset as an example (We will provide ASSIST dataset as soon as possible).

[//]: # (* **Note**: Concept dependency local map has been provided (see the instruction of dataset). The construction of concept dependency relation see subsection 5.1.2 in the paper. If you need, we would release this code.)

Go to the code directory:
```
cd RCD/RCD
```
Create two folders '/model' and '/result'.

Build exercise-concept correlation local map:
```
python build_k_e_graph.py
```

Build student-exercise interaction local map:
```
python build_u_e_graph.py
```
Train and test RCD model:
```
python main.py
```

**Note**: In subsection 4.3 (i.e., Extendable Diagnosis Layer) of the paper, Q_{e} in original MIRT represents exercise discrimination. We use a concept-related vector instead of discrimination as an implementation in the paper. RCD can be extended to the many forms of cognitive diagnosis.

## Dataset(RCD/data)
### junyi

log_data.json:
- Student exercising records.
- Source: https://github.com/bigdata-ustc/EduData

train_set.json
- Data file for training.

test_set.json
- Data file for testing.

graph/K_Directed.txt
- Prerequisite relation from concept dependency local map.
- Each line is a prerequisite relation from the concept dependency local map: precursor_concept_ID\t succeed_concept_ID.

graph/K_Undirected.txt
- Similarity relation from concept dependency local map.
- Each line is a similarity relation from concept dependency local map: concept_ID\t similar_concept_ID.

**Note**: Exercise-concept correlation local map and student-exercise interaction local map can be constructed by running build_k_e_graph.py and build_u_e_graph.py respectively.

### ASSIST

log_data.json:
- Student exercising records.

## Related Works
**Leveraging Transferable Knowledge Concept Graph Embedding for Cold-Start Cognitive Diagnosis (SIGIR'2023).** [[Paper](https://dl.acm.org/doi/10.1145/3539618.3591774)][[Code](https://github.com/WebGao/TechCD)][[Presentation Video](https://dl.acm.org/action/downloadSupplement?doi=10.1145%2F3539618.3591774&file=SIGIR23-fp1870.mp4)]

## BibTex
Please cite this paper if you use our codes. Thanks!

```
@inproceedings{gao2021rcd,
  title={RCD: Relation map driven cognitive diagnosis for intelligent education systems},
  author={Gao, Weibo and Liu, Qi and Huang, Zhenya and Yin, Yu and Bi, Haoyang and Wang, Mu-Chun and Ma, Jianhui and Wang, Shijin and Su, Yu},
  booktitle={Proceedings of the 44th international ACM SIGIR conference on research and development in information retrieval},
  pages={501--510},
  year={2021}
}
```

### Last Update Date: August 29, 2023