Reconciling inconsistent perspectives in collaborative GIS using multi-agent method

Reconciliation of inconsistent perspectives is an important issue for collaborative GIS and group decision making support systems. Although some methods, such as extended speech act theory (ESAT), have been used to handle this issue, some problems still remain unsolved such as how to define and process the negotiation workflow, how to handle the complex interactions among distributed users, and how to define the reconciliation objects. The multi-agent system, which has special features in adaptivity, autonomy, collaborative behavior, mobility and so on, has the potential to help solve these problems in collaborative computing, shared knowledge and rules, and complex communication, negotiation and voting. In this paper, multi-agent method is used to solve reconciliation of inconsistent perspectives in a real-time collaborative 3D environment. First, an application-based ontology is created to define the system-wide status including user profiles, user behaviors, system functions and status, and data and data sources. These statuses will be understandable by all the parties in the system. The reconciliation objects are also derived from the ontology. Second, multiple users´ negotiation process is simulated and is modeled based on the agent communication protocol from the foundation for intelligent physical agents (FIPA). Third, assistant agent and other agents are designed and developed to deploy the reconciliation process. A prototype system has been developed, which is structured into two layers: multi-agent system layer and shared 3D view layer. Using the prototype system, the geographically-dispersed users are aware of others´ operations, and can share the 3D GIS model, communicate, and even negotiate with each other and carry out common tasks.