Beyond the Clearinghouse

Agent-based mechanims for distributing geographic information services on the Internet.

Ming-Hsiang Tsou and Barbara P. Buttenfield, GIScience 2000: The First International Conference on Geographic Information Science, 2000

Abstract
The progress of information technology and the need for global distribution of geographic information is pushing the GIS community to distribute Geographic Information Services (GIServices) on the Internet. Currently, on-line GIService solutions are monolithic and platform-centric. Data sharing is becoming commonplace, but GIS operators and procedures remain isolated within vendor-disseminated software suites. Additionally, as data file size continues to increase, the time required to download data for client-side processing impedes modeling, environmental monitoring, and spatial decision support. The solution is not only that processors should be sent to large data clearinghouses, but that the approach to distributed processing should be entirely re-thought. An integrated network of GIS nodes can share the multiplicity of processing tasks by exchanging data or processes intelligently, according to the task, the speed of the node, and the network speed. This research proposes an agent-based communication mechanism to facilitate the dynamic integration of geospatial data, GIS programs, and modeling procedures in distributed network environments. The goal of agents is to reduce user work and information overload. In a distributed GIService environment, users interact with heterogeneous data models and different types of programs on different computer platforms. Agents help users to access distributed data objects and GIS components on heterogeneous GIS platforms by interpreting, filtering, and converting information automatically. An agent is an autonomous computer program that has specific functions and responds to specific events, based on pre-defined knowledge rules or user designated instructions. Three fundamental roles of agents are essential to distributed network environments: information finder/filter, information interpreter, and decision maker. An information finder and filter helps users to locate requested information and filter out unnecessary information needed for a specified task. The agent provides a number of choices and/or suggests alternative options if the requested information can not be discovered. An information interpreter conveys information during an information exchange. In distributed network environments, heterogeneous data models and systems can not communicate directly. An agent can bridge heterogeneous information systems and translate different data types and models for different data types and models for different GIS tasks. A decision maker makes decisions autonomously based on its embedded knowledge or on user-defined rules. An agent can collect and analyze information related to specific events, as for example in the migration of data objects and processing components between GIS nodes. The decision to move a data object to a different node for example can be based on the node characteristics. The agent makes an optimal decision based on the rational rules defined by users or other agents. In the latter case, this constitutes autonomous agent-to-agent communication. This paper presents a GIServices architecture based on agent mobility and functionality. The mobility of agents relates to the dynamics of network environments and metadata communicated between different machines. Stationary agents and mobile agents distinguish two types of functionality and responsibility for different tasks, such as communication between machines, fusion of GIS procedures from different vendors, and search for requested geodata objects. In the paper, geodata agents, processing component agents, and system-level machine agents are distinguished, along with their functionality in distributed GIServices. The agent-based GIServices architecture will be illustrated by the Unified Modeling Language (UML) to demonstrate the design of a high-level communication mechanism for heterogeneous GIServices. The paper will present specific examples to demonstrate how agent collaboration can provide a flexible and scalable architecture for distributing GIServices on the Internet.