This repository contains the code used for the experiments in the paper "In-Context Meta-Learning vs. Semantic Score-Based Similarity: A Comparative Study in Arabic Short Answer Grading" by Fateen et. al. (2023).
To install requirements:
- Create a conda environemnt with specified python version
conda create -n sss-vs-incml python=3.9
2.Install the required libraries by running
pip install -r requirements.txt
The code is organized as follows:
data/
: contains the data used in the experiments.sss/
: contains the code for the SSS model. The files can be run in the order defined by the file names.incml/
: contains the code for the INCML model. The files can be run in the order defined by the file names.prompt_classify.py
: the code for the prompt classification model.
The data used in the experiments was collected from the Arabic ASAG dataset
We divided the data per prompt type and divided them into train/test sets. The data can be found in the data/csv
folder.
If you use this code in your research, please cite the following paper:
@inproceedings{fateen2023incontext,
title={In-Context Meta-Learning vs. Semantic Score-Based Similarity: A Comparative Study in Arabic Short Answer Grading},
author={Fateen, Menna and Mine, Tsunenori},
year={2023}
}