RoBERTa model for detecting framing elements in tweets related to social movements.

Multilabel classification of 7 categories:

- problem identification (diagnostic core framing task)
- blame attribution (diagnostic core framing task)
- proposing solutions (prognostic core framing task)
- discussing tactics (prognostic core framing task)
- expressing solidarity / celebrating a movement (prognostic core framing task)
- counterframing / challenging the opposite sides' arguments (prognostic core framing task)
- motivational / calls to action (motivational core framing task)

The model is trained on Twitter data from three issue areas: guns, LGBTQ rights, and immigration. 
All data is in English and written in 2018-2019.

Paper: https://journalqd.org/article/view/5896