Exploring the Effectiveness of Attention Manipulations on Operator Performance in C2 Task: Computational Modeling Approach.

摘要

The goal of the project was to develop computational models that helped assess operator performance during cyberoperations. Cyberoperation agents with good user models can be deployed to simultaneously monitor the effects of cyberoperations on several users. This in turn increases the trust in and success of these operations. A key obstacle to these user models was the limitations of the existing inference algorithms used to create them. These algorithms did not scale well in domains that extended over time, contained uncertainty over identity, included large numbers of entities and involved reasoning about the beliefs of other agents. Cyberoperations have all these properties and thus there was a large gap between the needs of these operations and the abilities of existing inference mechanisms. In this project we developed inference mechanisms that made significant progress towards overcoming most of these obstacles.