Model Card for uvegesistvan/wildmann_german_proposal_1

Model Overview

This model is a multi-class emotion classifier trained to identify nine distinct emotional states in text. The classes and their corresponding labels are as follows:

Anger (0)
Fear (1)
Disgust (2)
Sadness (3)
Joy (4)
Enthusiasm (5)
Hope (6)
Pride (7)
No emotion (8)

Dataset and Preprocessing

The dataset underwent undersampling to balance the two most frequent classes ("Anger" and "No emotion") with the others. This adjustment aimed to mitigate class imbalance and improve model performance across all labels.

Evaluation Metrics

The model was evaluated using precision, recall, F1-score, and support for each class. Below are the detailed metrics:

Class	Precision	Recall	F1-Score	Support
Anger (0)	0.62	0.63	0.62	777
Fear (1)	0.49	0.59	0.54	317
Disgust (2)	0.61	0.56	0.59	105
Sadness (3)	0.65	0.61	0.63	333
Joy (4)	0.65	0.71	0.68	427
Enthusiasm (5)	0.45	0.42	0.44	544
Hope (6)	0.53	0.57	0.55	777
Pride (7)	0.47	0.57	0.52	354
No emotion (8)	0.46	0.36	0.41	777

Overall Metrics

Accuracy: 0.54
Macro Average: Precision = 0.55, Recall = 0.56, F1-Score = 0.55
Weighted Average: Precision = 0.54, Recall = 0.54, F1-Score = 0.54

Performance Insights

The model achieves reasonable accuracy and F1-scores across most classes. However, classes like "Fear" (1) and "No emotion" (8) exhibit lower performance, which may stem from either insufficient training samples or ambiguous cases in the dataset.

Model Usage

Applications

Emotion classification in text-based datasets.
Analyzing emotional tone in social media, reviews, or other text corpora.

Limitations

Performance varies across classes, with some (e.g., "Fear" and "No emotion") showing lower recall.
The model may not generalize well to domains outside the training data.

Ethical Considerations

The model's predictions might not always align with human interpretations of emotions, particularly in ambiguous or context-dependent cases. Misclassification could lead to inappropriate conclusions if used in sensitive applications (e.g., mental health monitoring).