K-EmoPhone: A Mobile and Wearable Dataset with In-Situ Emotion, Stress, and Attention Labels

Area of Science:

• Affective computing
• Human-computer interaction
• Mobile sensing

Background:

• Mobile and wearable sensors are increasingly used for mental well-being analysis.
• A significant gap exists in open-access, real-world datasets with affective and cognitive state labels.
• This limitation hinders progress in affective computing and human-computer interaction research.

Purpose of the Study:

• To introduce K-EmoPhone, a novel multimodal dataset.
• To provide a rich resource for studying affective and cognitive states in real-world settings.
• To facilitate advancements in emotion intelligence and attention management technologies.

Main Methods:

• Collected data from 77 students over seven days using commercial mobile and wearable sensors.
• Acquired peripheral physiological signals, mobility data, and smartphone context/interaction data.
• Utilized the experience sampling method to gather 5,582 self-reported affect states (emotions, stress, attention).

Main Results:

• The K-EmoPhone dataset comprises multimodal data including physiological signals, mobility, context, and self-reported affect.
• It captures real-world affective and cognitive states over an extended period.
• The dataset is designed for research in affective computing and attention management.

Conclusions:

• The K-EmoPhone dataset addresses the need for real-world affective and cognitive data.
• It is expected to significantly contribute to the development of emotion intelligence and attention management tools.
• This resource will accelerate research in mobile sensing for mental well-being and human-computer interaction.