MAVERICK: A Synthetic Murder Mystery Network Dataset to Support Sensemaking Research

Jenkins1, M., Bisantz2, A., Nagi3, R., and Llinas2, J.

Presented at the International Conference on Applied Human Factors and Ergonomics (AHFE), Las Vegas, NV (July 2015).

The MAVERICK dataset was created to support a series of empirical studies looking at the effectiveness of network visualizations intended to support information foraging and human sensemaking within the domain of counterinsurgency intelligence analysis. This synthetic dataset is structured as a forensic mystery with the central goal of solving a fictional murder. The dataset includes 181 text-based reports, with additional media included with some messages as attachments, collected from various sources of varying reliability. The reports are framed as being collected from the perspective of a reporter investigating the murder through interviews with suspects and observations taken at the site the murder. The dataset includes intentional and unintentional deception along with calculated source reliabilities based on available evidence.

The dataset is dynamic in nature, as the information in the dataset evolves and expands over a simulated period of time. This is done to both to simulate a real-world scenario, and to allow for evolutionary tasks and experiments to be performed using the dataset. The dataset is designed to be complex enough to simulate a real-world, while remaining accessible to individuals without experience in a specific domain of interest. This meant that it had to be on a topic that did not require prior domain knowledge to understand available information or to understand what strategies should be applied during analysis of the dataset. The solution to these challenges was the development of a fictional murder mystery. The plot involves a murder that took place over the course of a weekend with several possible suspects at a large private estate. This scenario allowed for a great deal of complexity; however, it was also a subject matter that could be easily understood by participants without prerequisite domain experience.

Charles River Analytics
University of Buffalo, The State Univeristy of New York
University of Illinois at Urbana-Champaign

For More Information

To learn more or request a copy of a paper (if available), contact Michael Jenkins.

(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)