Emotion Classifier

Model Details

• Model Type: Fine-tuned transformer-based text classification model
• Base Model: RoBERTa
• Training Dataset: MELD (Multimodal EmotionLines Dataset)
• Number of Parameters: ~125M
• Sequence Length: 128 tokens
• Training Approach: Fine-tuned with cross-validation

Intended Use

• Sentiment analysis in customer feedback
• Emotion detection in conversations
• User experience research
• Content moderation
• Game development for adaptive emotional responses

Limitations

• The model was trained on scripted dialogues from TV shows, which may not fully represent natural conversations
• Short texts may be harder to classify accurately
• Cultural and contextual nuances might not be captured
• The model may reflect biases present in the training data

Performance

• Accuracy: [Insert your model's accuracy]
• F1 Score: [Insert your model's F1 score]
• Training Dataset Size: ~13,000 utterances

API Usage

Ethical Considerations

• Do not use this model to manipulate people's emotions
• Be transparent when using emotion detection in user-facing applications
• Do not make high-stakes decisions based solely on this model's outputs
• Consider privacy implications when analyzing personal communications

Citation

@inproceedings{poria-etal-2019-meld,
    title = "{MELD}: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations",
    author = "Poria, Soujanya  and
      Hazarika, Devamanyu  and
      Majumder, Navonil  and
      Naik, Gautam  and
      Cambria, Erik  and
      Mihalcea, Rada",
    booktitle = "Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    pages = "527--536"
}