- torch 1.7.1+cu110
- transformers 4.5.1
The raw data we used is available here, and our proposed data is available here
- Description to directory of our processed data
.
|-processed_data # processed data in the format of pickle
|-split_data # splits of train, dev and test
|rawdata.txt
mkdir log
cd ./src/
- train (Other parameters have been set as default.)
CUDA_VISIBLE_DEVICES=1 python train.py --task_name ece_task --training 1 --debug 0 --hidden_size 768
- Inference
CUDA_VISIBLE_DEVICES=1 python train.py --task_name ece_task --training 0 --debug 0 --hidden_size 768 --model_name model_name
Please cite our paper as
@inproceedings{cui-etal-2022-event,
title = "Event Causality Extraction with Event Argument Correlations",
author = "Cui, Shiyao and
Sheng, Jiawei and
Cong, Xin and
Li, Quangang and
Liu, Tingwen and
Shi, Jinqiao",
booktitle = "Proceedings of the 29th International Conference on Computational Linguistics",
month = oct,
year = "2022",
address = "Gyeongju, Republic of Korea",
publisher = "International Committee on Computational Linguistics",
url = "https://aclanthology.org/2022.coling-1.201",
pages = "2300--2312",
abstract = "Event Causality Identification (ECI), which aims to detect whether a causality relation exists between two given textual events, is an important task for event causality understanding. However, the ECI task ignores crucial event structure and cause-effect causality component information, making it struggle for downstream applications. In this paper, we introduce a novel task, namely Event Causality Extraction (ECE), aiming to extract the cause-effect event causality pairs with their structured event information from plain texts. The ECE task is more challenging since each event can contain multiple event arguments, posing fine-grained correlations between events to decide the cause-effect event pair. Hence, we propose a method with a dual grid tagging scheme to capture the intra- and inter-event argument correlations for ECE. Further, we devise a event type-enhanced model architecture to realize the dual grid tagging scheme. Experiments demonstrate the effectiveness of our method, and extensive analyses point out several future directions for ECE.",
}
and our used dataset as
@misc{tianchi2021EventCausal,
title={CCKS2021 The Dataset for Financial Event and Causal Relation Extraction},
url={https://tianchi.aliyun.com/dataset/dataDetail?dataId=110901},
author={Tianchi},
year={2021},
}