Awesome

FewFC

• Few-shot Financial Chinese event extraction datase

Important

• In order to accelerate the research on few-shot event extraction, incremental event extraction and low resources event extraction, we introduce FewFC, a new Chinese Financial sentence-level event extraction
• dataset constructed from news reports on the Internet and announcements issued by listed companies. This data set is also used for CCKS TASK 3 evaluation.

Details

The data set contains 10 financial field event types and 8982 sentences.

Event_Definition.pdf

• Explains the definition of event types and event roles

Rearranged Dataset

train_base.json

• Training set from source domain with 5 event types.

test_base.json

• Test set from source domain.

train_trans.json

• Training set from target domain.

test_trans.json

• Test set from target domain.

Original Dataset

• Data divided according to the CCKS evaluation

train_base.json

• Training set from source domain in Phase I Evaluation. Same as train_base.json in rearranged dataset.

train_trans_2_type.json

• Training set from target domain with only 2 event types in Phase I Evaluation. Part of the train_trans.json in rearranged dataset.

dev_base.json

• Development set from source domain in Phase I Evaluation. Same as test_base.json in rearranged dateset.

dev_trans.json

• Developement set from target domain with only 2 event types in Phase I Evaluation. Part of the test_trans.json in rearranged dataset.

train_trans_5_type.json

• Training set from target domain with all 5 event types in Phase II Evaluation. Same as train_trans.json in rearranged dataset.

test_trans.json

• Test set from target domain with all 5 event types. Part of the test_trans.json in rearranged dataset.

Partial Dataset

• Data division used in the paper “What the role is vs. What plays the role:Semi-supervised Event Argument Extraction via Dual Question Answering”

Cite

If you use the dataset or the code, please cite this paper:

@inproceedings{Yang2021,
  author    = {Yang Zhou and
               Yubo Chen and
               Jun Zhao and
               Yin Wu and
               Jiexin Xu and
               Jinlong Li
               },
  title     = {What the role is vs. What plays the role: Semi-supervised Event Argument Extraction via Dual Question Answering},
  booktitle = {Proceedings of AAAI-21},
  publisher = {{AAAI} Press},
  year      = {2021},
}

License

Copyright: This data set is released by the nlp group of the Institute of Automation of the Chinese Academy of Sciences, collected by China Merchants Bank, licensed under CC BY-SA 4.0.