CausalFSED

• An implementation for EMNLP 2021 short paper Honey or Poison? Solving the Trigger Curse in Few-shot Event Detection via Causal Intervention

Quick links

• Environment
• Dataset
• Model Training
• Model Evaluation

Environment

conda create -n causalfsed python=3.7.3
conda activate causalfsed
pip install -r requirements.txt

Dataset

Custom dataset can be putted in data folder:

data/custom
├── sentence.json
├── test.json
├── train.json
├── dev.json
└── getquery.py

sentence.json is data file, and the format is List and each element is a Dict. Each Dict contains words and event fields, in which text is the list of tokens, and event is the list of event mentions.

[
    {
        "words": [token1,token2,...],
        "event": [
            [
                [eventtype1, {"text": trigger word1, "start": startindex1, "end": endindex2],
                [eventtype2, {"text": trigger word2, "start": startindex2, "end": endindex2],
                ...
            ]
    },
    ...
]

train/dev/test.json record the index of sentence for train/dev/test set, and the format is Dict. The key of Dict is event type and the value of Dict is a List which contains the index of sentence in sentecen.json.

train.json
{
    eventtype1: [sentenceindex1, sentenceindex2, ...], 
    eventtype2: [sentenceindex1, sentenceindex2, ...],
    ...
}

The file getquery.py is used to obtain ambigous triggers for dev set which are used in training.

cd data/custom
python getquery.py

For ACE2005, KBP2017 and MAVEN dataset, the files data/ace2005/acepreprocess.py, data/maven/mavenpreprocess.py and data/kbp2017/kbppreprocess.py can be used to convert some other format to this format.

Model Training

run:

python main.py -model FSCausal -metric proto -dataset ace -cuda 0

• -model can be FSBase and FSCausal
• -metric can be proto and relation
• -dataset can be ace, kbp, maven and custom
• -cudaid is the gpu id.

where proto is Prototypical Network and relation is Relation Network.

The model checkpoint will be saved in tmp/dataset_name/...

Model Evaluation

just add -t:

python main.py -model FSCausal -metric proto -dataset ace -cuda 0 -t