Information extraction in emergency management missions: an adaptive multi-agent approach

Abstract

With increasing demands for autonomous agents to work alongside humans in emergency management response (EMR), considerations of translations of human to machine language (and the converse) are timely. We present a prototype where the translation is dealt with by restricting communications to occur through a form of controlled natural language (CNL) (Fuchs and Schwitter, 1995). The prototype is new in that it allows for communications between both physical and virtual autonomous agents, agents are assigned different levels of autonomy, and it includes a level of information hiding that allows for information to be passed to relevant agents, whilst keeping those (humans) involved anonymous. A real-life mission is then used to exemplify how information is retrieved and communicated in the prototype. Finally, some usability experimental results are presented.