Geospatial information is a fundamental component of many crisis management activities. However, current geospatial technologies do not support work by crisis management personnel, most of whom are not technology specialists—a key impediment is that the technologies require the user to learn the system's language. In addition, geospatial technologies are not ‘collaboration friendly’—they impede rather than facilitate group work. In this paper we address both issues by presenting (1) a theoretical framework for understanding the roles of visual mediation in map-supported human–human dialogues, and (2) a computational approach for enabling such roles in collaborative spatial decisionmaking contexts. Building upon our initial implementation of a map-mediated collaborative environment, the DAVE_G system [a natural, multimodal, dialogue-enabled interface to geographical information systems (GIS)], we model human–GIS and human–GIS–human dialogues as complex visual-cognitive signification processes in which maps become dynamic facilitators. Using a scenario simulating two crisis managers dealing with a major nuclear release event, we demonstrate how visual display (in DAVE_G) actively mediates human–human dialogue directed to situation assessment and action planning in real applications.

Crises are often complex, geographical scale problems that require professionals to work in teams while dealing with a large amount of geographic information for decision making. However, current geospatial technologies do not directly support group entities working with geographic information–they impede rather than facilitate human-human collaborations and communication. Towards the goal of making GIS “collaboration-friendly,” this paper explores the potentials of extending distributed GIS with groupware and intelligent communication agents to support geo-collaborative crisis management by teams. Members of such a team are often geographically distributed and play different roles. In addition to the architectural choice, special attention was given to the computational approach to enable collaborative geographic information dialogues in spatial decision-making contexts. Collaboration requires representation and reasoning on a team mental model, which must be constructed from dialogue contexts and shared knowledge. An implementation of an intelligent, multimodal, multi-user geographic information environment, called GCCM_Connect, is presented as a proof-of-concept for the proposed architecture.