Event-Causality-Extraction

Description:

The code and prompts from the paper Event Causality Is Key to Computational Story Understanding, Accepted at NAACL 2024.

The paper proposes a prompt for extracting event causality from free form story text with LLMs, and show that the extracted event causality improves performence on downstram tasks.

Requirement:

Python version is 3.9

requirements:

numpy                     1.23.5
openai                    0.27.0
openssl                   1.1.1w
pandas                    2.0.2
torch                     2.0.1
tokenizers                0.13.3
transformers              4.28.0
sacrebleu                 2.3.1                    
scikit-learn              1.2.2
rouge                     1.0.1

You can install all requirements with the command:

pip install -r requirements.txt

Datasets

Event Causality Extraction

COPES

The COPES dataset can be downloaded from: https://github.com/HKUST-KnowComp/COLA/tree/master

GLUCOSE

The GLUCOSE dataset can be downloaded from: https://github.com/ElementalCognition/glucose

Downstream Tasks:

Story Evaluation

The dataset has been downloaded and save in the story_eval/datasets/ folder

Video-Text Alignment

To be filled

Prompt

The prompt for event causality extraction is stored in prompt.txt

Run event causality extraction on COPES:

The code is in COPES/event_causality_extraction_copes.py

event_causality_extraction_copes.py --save_dict {output file} --model {model_name} --input_data {path to COPES.json} --data_split {path to data split} --prompt {path to prompt}

Run event causality extraction on GLUCOSE

The code is in GLUCOSE/event_causality_extraction_glucose.py

event_causality_extraction_glucose.py --save_dict {output file} --input_data {path to input data file} --prompt {path to prompt}

Run event causality extraction on OpenMEVA datasets

The code is in story_eval/OpenAI_API_OpenMEVA_EN.py

OpenAI_API_OpenMEVA_EN.py -d ROC --seed 2 --gpt_model gpt4

Run story evaluation experiments on OpenMEVA datasets

The code is in story_eval/OpenAI_API_score.py

OpenAI_API_score.py --prompt_type orig -d ROC --seed 2 --gpt_model gpt4

Citation

If you find our paper is useful for your research and applications, please cite using this BibTeX:

@inproceedings{Sun2023EventCI,
  title={Event Causality Is Key to Computational Story Understanding},
  author={Yidan Sun and Qin Chao and Boyang Albert Li},
  booktitle={The Annual Conference of the North American Chapter of the Association for Computational Linguistics},
  year={2024},
  url={https://arxiv.org/abs/2311.09648}
}