Between May 2021 and March 2022, video data from 84 participants in 28 groups was collected (the exact description of this data can be found in the article cited uder related identifiers). For the present analysis, facial action units were extracted from the videos and used for a machine learning based classification of conversation topic. Additionally, the videos were rated by 19 individuals, between May and June 2023.

This repository contains (1) the maschine learning dataset with individuals' facial action units and code to classify group discussion topics as a Python script with instructions and (2) Human rating data for third-person, video-based classification and analysis code as an R-script.

The present data was collected as part of the Collective Appetite Project in the Cluster for the Advanced Study of Collective Behavior at the University of Konstanz. It belongs to the article cited under related identifiers.