EQGG: Automatic Question Group Generation

Abstract

Question generation (QG) task plays a crucial role in adaptive learning. While significant QG performance advancements are reported, the existing QG studies are still far from practical usage. One point that needs strengthening is to consider the generation of question group , which remains untouched. For forming a question group, intra-factors among generated questions should be considered. This paper proposes a two-stage framework by combining neural language models and genetic algorithms for addressing the issue of question group generation.

Components

There are some components spread across multiple repositories:

• QGG-RACE dataset
• qgg-demo
• qgg-utils

Negative Label Loss, Genetic algorithm and Evaluation scorer

Environment

Require

• OS:ubuntu 16.04+
• RAM:24GB
• GPU: CUDA device with 12GB VRAM

Setup

Please install pytorch>=1.7.1,<=1.9.0 manually first

PyTorch install: https://pytorch.org/get-started/locally/

sudo apt install unzip
pip install -Ur requirements.txt
python -c "import stanza;stanza.download('en')"
python init_dataset.py
python setup_scorer.py

Training

python train_xxx.py -m'message to note for this training'

Citation

If your research references the relevant content, please cite:

@ARTICLE{10609322,
  author={Huang, Po-Chun and Chan, Ying-Hong and Yang, Ching-Yu and Chen, Hung-Yuan and Fan, Yao-Chung},
  journal={IEEE Transactions on Learning Technologies}, 
  title={EQGG: Automatic Question Group Generation}, 
  year={2024},
  volume={},
  number={},
  pages={1-14},
  keywords={Task analysis;Context modeling;Question generation;Training;Redundancy;Fans;Employment;Neural Question Generation;Natural Language Generation;Reading Comprehension Testing},
  doi={10.1109/TLT.2024.3430225}
}