BIRAFFE2, a multimodal dataset for emotion-based personalization in rich affective game environments

Area of Science:

• Affective Computing
• Human-Computer Interaction
• Biometric Signal Processing

Background:

• Current emotion prediction models using physiological data lack robustness.
• Personalization and contextual information are crucial for improving model effectiveness.
• A need exists for comprehensive datasets to facilitate personalized emotion recognition.

Purpose of the Study:

• To introduce the Bio-Reactions and Faces for Emotion-based Personalization for AI Systems (BIRAFFE2) dataset.
• To address the limitations of existing datasets in affective computing.
• To enable the development of more robust and personalized emotion prediction models.

Main Methods:

• Collected multimodal data from 102 participants, including physiological signals (accelerometer, ECG, EDA).
• Acquired facial expression data and contextual information from an affective gaming session.
• Incorporated personality and game engagement questionnaires to capture individual differences.

Main Results:

• Validated the correctness of collected contextual data from the gaming environment.
• Identified significant relationships between participant personality traits and their emotions.
• Revealed correlations between personality and physiological signals, supporting personalization.

Conclusions:

• The BIRAFFE2 dataset provides a rich resource for advancing personalized emotion recognition.
• Integrating contextual and personality data enhances the potential for accurate emotion prediction.
• This dataset facilitates research in affective computing, particularly for AI systems.