roberta_finetuned_multilabel_unions

Multi-Label RoBERTa Classifier for Union Communication Frames

This model is a multi-label classifier fine-tuned on U.S. labor union Facebook posts. It predicts the following communication frames:

• Diagnostic
• Prognostic
• Motivational
• Community
• Engagement
• Political Endorsement

The model was trained to automatically label union posts according to these categories, supporting research on social media discourse around labor unions. The predicted outcome is a list of 6 binary labels of the kind [0, 0, 1, 0, 0, 0]. Each binary label correspond to a discourse frame in the order presented above.

More details can be found in the paper "Framing Unionization on Facebook: Communication around Representation Elections in the United States".

---

Model Details

• Base model: RoBERTa-base
• Task: Multi-label text classification
• Domain: Labor union Facebook posts
• Framework: Simple Transformers

---

Usage

from simpletransformers.classification import MultiLabelClassificationModel
import torch
import pandas as pd

def main():

    cuda_available = torch.cuda.is_available()

    model_args = {
        "process_count": 1,  # disables multiprocessing
    }

    model = MultiLabelClassificationModel(
        "roberta",
        "ariannap22/roberta_finetuned_multilabel_unions",
        args=model_args,
        use_cuda=cuda_available
    )

    # Example inference
    texts = [
        "We successfully won the contract vote!",
        "Join us for the upcoming rally to support labor rights."
    ]

    predictions, raw_outputs = model.predict(texts)

    print(predictions)

if __name__ == "__main__":
    main()