language: german
Welcome to ParlBERT-Topic-German!
🏷 Model description
This model was trained on ~10k manually annotated political interpellations (📚 Breunig/ Schnatterer 2019) of comparative agenda topics to classify text into one of twenty labels (annotation codebook).
🗃 Dataset
| party | speeches | tokens |
|---|---|---|
| CDU/CSU | 7,635 | 4,862,654 |
| SPD | 5,321 | 3,158,315 |
| AfD | 3,465 | 1,844,707 |
| FDP | 3,067 | 1,593,108 |
| The Greens | 2,866 | 1,522,305 |
| The Left | 2,671 | 1,394,089 |
| cross-bencher | 200 | 86,170 |
🏃🏼♂️Model training
ParlBERT-Topic was fine-tuned on a domain adapted model for topic modeling with interpellations dataset from the Comparative Agendas Project (mlm_probability=.15). We used the HuggingFace trainer with the following hyperparameters.
🤖 Use
from transformers import pipeline
pipeline_classification_topics = pipeline("text-classification", model="chkla/parlbert-topics-german", tokenizer="bert-base-german-cased", return_all_scores=False, device=0)
text = "Sachgebiet Ausschließliche Gesetzgebungskompetenz des Bundes über die Zusammenarbeit des Bundes und der Länder zum Schutze der freiheitlichen demokratischen Grundordnung, des Bestandes und der Sicherheit des Bundes oder eines Landes Wir fragen die Bundesregierung"
pipeline_classification_topics(text) # Government
📊 Evaluation
The model was evaluated on an evaluation set (20%):
| Label | F1 | support |
|---|---|---|
| International | 80.0 | 1,126 |
| Defense | 85.0 | 1,099 |
| Government | 71.3 | 989 |
| International | 76.5 | 978 |
| International | 76.6 | 845 |
| International | 86.0 | 800 |
| International | 67.1 | 0.8021 |
| International | 78.6 | 0.8021 |
| International | 78.2 | 0.8021 |
| International | 64.4 | 0.8021 |
| International | 81.0 | 0.8021 |
| International | 69.1 | 0.8021 |
| International | 62.8 | 0.8021 |
| International | 76.3 | 0.8021 |
| International | 49.2 | 0.8021 |
| International | 63.0 | 0.8021 |
| International | 71.6 | 0.8021 |
| International | 79.6 | 0.8021 |
| International | 61.5 | 0.8021 |
| International | 45.4 | 0.8021 |
⚠️ Intended Uses & Potential Limitations
The model can only be a starting point to dive into the exciting field of policy topic classification in political texts. But be aware. Models are often highly topic dependent. Therefore, the model may perform less well on different topics and text types not included in the training set.
👥 Cite
@article{klamm2022frameast,
title={FrameASt: A Framework for Second-level Agenda Setting in Parliamentary Debates through the Lense of Comparative Agenda Topics},
author={Klamm, Christopher and Rehbein, Ines and Ponzetto, Simone},
journal={ParlaCLARIN III at LREC2022},
year={2022}
}
🐦 Twitter: @chklamm