Model Card for uvegesistvan/wildmann_german_proposal_2b_german_to_slovak

Model Overview

This model is a multi-class emotion classifier trained on German-to-Slovak machine-translated text data. It identifies nine distinct emotional states in text. The model utilizes a diverse dataset, incorporating both synthetic and original German sentences translated into Slovak, highlighting its ability to generalize across linguistic variations introduced by machine translation.

Emotion Classes

The model classifies the following emotional states:

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

Dataset and Preprocessing

The dataset consists of German text machine-translated into Slovak and annotated for emotional content. It includes both synthetic and original sentences to enhance diversity. Preprocessing involved:

Undersampling of overrepresented classes, such as "No emotion" and "Anger," to ensure balanced training across all labels.

Evaluation Metrics

The model's performance was evaluated using precision, recall, F1-score, and accuracy metrics. Detailed results are as follows:

Class	Precision	Recall	F1-Score	Support
Anger (0)	0.54	0.58	0.56	777
Fear (1)	0.84	0.78	0.81	776
Disgust (2)	0.93	0.94	0.93	776
Sadness (3)	0.85	0.84	0.84	775
Joy (4)	0.82	0.80	0.81	777
Enthusiasm (5)	0.62	0.64	0.63	776
Hope (6)	0.53	0.54	0.54	777
Pride (7)	0.74	0.79	0.77	776
No emotion (8)	0.67	0.63	0.65	1553

Overall Metrics

Accuracy: 0.72
Macro Average: Precision = 0.73, Recall = 0.73, F1-Score = 0.73
Weighted Average: Precision = 0.72, Recall = 0.72, F1-Score = 0.72

Performance Insights

The model demonstrates robust performance in detecting "Fear" and "Disgust," while "Hope" and "Enthusiasm" show slightly lower performance due to subtleties in emotional expression and potential translation noise. These results reflect the complexities of training on machine-translated text.

Model Usage

Applications

Emotion analysis of German texts translated into Slovak for social research or sentiment tracking.
Research on cross-linguistic emotion classification in multilingual datasets.
Sentiment analysis for Slovak-language customer feedback derived from German text.

Limitations

The model's performance depends on the quality of the machine-translated text. Translation errors or ambiguities may impact classification accuracy.
Subtle emotional expressions may be misclassified due to linguistic nuances being lost in translation.

Ethical Considerations

The use of machine-translated datasets introduces the possibility of biases or inaccuracies caused by the loss of cultural and linguistic subtleties during translation. Users should carefully evaluate the model's performance before applying it in sensitive contexts such as mental health, social studies, or customer sentiment analysis.

Citation

For further information, visit: uvegesistvan/wildmann_german_proposal_2b_german_to_slovak