ABSTRACT In China, with the rise of multi-child families, enhancing interactions between siblings to promote their mutual learning has been receiving widespread attention. Among the multitude of educational toys, robotic programming is becoming increasingly popular. Existing datasets for children education predominantly focus on interactions between unrelated children, between parents and children, or between children and electronic devices. However, these datasets do not encompass the unique aspects inherent in sibling interactions, such as their unspoken mutual understanding. To bridge this gap, we introduce a sibling interaction analysis dataset (SIBD) based on video recordings of 48 children (24 sibling pairs) from Chinese households. SIBD contains manually annotated labels and automated extracted features, giving a clear view into sibling interactions during learning processes. Beyond the programming robot scenario we employed, the dataset can be extended to encompass other contexts such as block building, science experiment kits, and similar collaborative activities. The statistics of the dataset identified potential relationships among the labels, indicating that siblings with a more positive interactive atmosphere and timely responses could be more likely to stay focused on tasks. SIBD can serve as a foundation for developing targeted interventions that improve interactions during sibling learning. Moreover, the insights derived from this study, while grounded in the Chinese context, may offer preliminary implications for understanding sibling interactions within similar cultural contexts.
Read full abstract