Presentation
Collective Anomaly Detection: A Method for Using Content-Free Communications to Predict Performance
DescriptionTeam dynamics can be understood through content-free communication analysis methods that measure patterns of communication and flow of information. Rolling Recurrence Quantification Analysis (RQA), for instance, quantifies dimensions of team interaction and results in continuous time series. However, interpreting these outputs with other continuous data streams to derive team-level measures can be difficult. Our approach, Collective Anomaly Detection (CAD), provides a framework for analyzing any continuous time series to identify off-nominal periods, such as communication associated with poor performance. CAD can incorporate many different types of data streams, like those generated by RQA, into a holistic and potentially real time analysis for interpreting complex team interactions. While this method was developed on communication data, CAD is theoretically generalizable to many different data types and task environments.
Event Type
Lecture
TimeWednesday, September 11th9:45am - 10:05am MST
LocationFLW Salon J
Communications