Vax-Culture

数据集获取

git clone https://github.com/mrzarei5/Vax-Culture.git

curl -L -o repo.zip https://github.com/mrzarei5/Vax-Culture/archive/refs/heads/main.zip
unzip repo.zip

源站 README 摘录（使用方式）

Requirements

Installation

Create a conda environment and install dependencies:

conda create name Vax_Culture file requirements.txt
conda activate Vax_Culture

Dataset

1. Copy the contents of Vax-Culture dataset directory (link) to filelists.

2. Pull the tweets using their provided unique identifiers from Twitter and append their texts to filelists/Vax_Culture.csv under the column name tweet_text. (Any subset of the tweets in Vax_Culture.csv can be dropped. Just remember not to alter the header of columns.)

3. Create train, validation and test subsets by running filelists/prepare_datasets.py:

python prepare_datasets.py

Classification Tasks

Run classification.py with the desired model and problem task:

python classification.py model_name Bert_base problem communicated_message

Parameter model_name can be selected from Bertweet, Bert_base, Bert_large, Roberta_base and Roberta_large for any of the four classification problem tasks.
Parameter problem can be selected from any of the values: inaccurate_or_misleadning for misleading or inaccurate information detection task, communicated_message for communicated message prediction task, multilabel_criticism for subjects of criticism prediction task, multilabel_support for subjects of support/promote prediction task.

Text Generation Task

Run generation.py with the desired model:

python generation.py model_name Bart_large problem gen

Parameter model_name can be Bart_large or T5_large.

数据加载示例（表格/文本类）

import pandas as pd, glob, os

files = (glob.glob(os.path.join(path, "**", "*.csv"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.tsv"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.xlsx"), recursive=True))
print("数据文件:", files)
df = pd.read_csv(files[0])
print(df.shape); print(df.columns.tolist()); print(df.head(3))

完整仓库：github.com/mrzarei5/Vax-Culture