Digital Social Dynamics

Analysis of Social Embedding

Humans are social creatures and connection to others is relevant for our well-being. The quantitative study of social interactions remains challenging due to the lack of scalable, privacy-preserving data that capture real-world interactions. We address this limitation with our data donation platform Dona, which allows people to donate their de-identified meta-data (e.g., timestamps, message length – no content!) from Facebook, Instagram, Whatsapp and iMessages.

Using this data, we apply time-series analyses and machine learning methods to examine how digital social interactions evolve over time and to identify patterns associated with different psychological characteristics. Among other topics, we study how major life events, such as personal or collective trauma (e.g., war) affect how people interact with others over time.

The analysis of human behavior data is methodologically challenging due to inherent burstiness and its dynamics across multiple temporal scales. By developing approaches that account for these properties, we aim to advance analytical tools that can be applied to similarly complex, temporally structured data in other domains.

Interested in working together?

With both analytical expertise and Dona's capability of enriching ongoing or past data collection efforts and studies with retrospective data, options for collaboration are manifold. If you are interested, please reach out to Florian Martin or Dr. Olya Hakobyan.

From 01.12.2020 – 30.11.2024, this research was funded by the "Empathische Künstliche Intelligenz" (EKI) grant, FKZ 01IS20046.