Agent Technology Applied In Map Generalization

This research aims to the status what the intelligence and automation of map generalization is not high and the engineering project of land-use data generalization requires. Utilizing the agent technology models and analyzes the process of map generalization so as to improve the map generalization’s automation and intelligence, and conducts a case study for land-use patch in order to enhance the efficiency of land-use data generalization. The content of the full text includes the following:1. This research analyzes the status of map generalization in several ways under the digital technology, including concepts, types, applications of expert system and agent technology, land-use data generalization, and so on. Then presents a viewpoint: Map generalization is a process of task decomposing and solving, that is dividing the whole map generalization into a series of different granularity size and strong independent subtasks, finishes the whole task of map generalization by solving each subtasks, and in the procedure of solving the map generalization’s task accompanied with human or computer’s intelligent behavior. The content of map generalization is reducing the spatial, semantics and temporal resolution of self-and contextual characteristics under the conditions of keeping precision. The map generalization is a procedure which involves in analyzing, decision-making and execution, quality evolution and trace tactic.2. This research applies the agent technologies to model the map generalization’s procedure, proposes a viewpoint the multi-agent system for map generalization should include five basic parts:task decomposition, task coordination, division of agent roles, individual agent operation process and agent communication. Divides the map generalization’s task into four levels in spatial terms:cartographic region, group objects, individual object and part of object, and according this to conduct agent role division, including map region agent at macro level, group map objects agent at meso level, map object agent at micro level, part of map object agent at part-micro level. This paper puts forward the corresponding relationship between task coordination in multi-agent system and process control in sequence of disposing map generalization’s task and map objects, proposes the basic components of individual agent and the primary steps in its life cycle of an agent in multi-agent system for map generalization, puts forward the blackboard model of agent communication in multi-agent system for map generalization.3. This research describes and detects the self-and contextual characteristics of the map object agent in multi-agent system for map generalization:Conducts expression and recognition of the self-characteristics of map object agent by dividing three characteristics including space, semantics and tense in content and distinguishing three kinds containing point, polyline and polygon based on geometric dimension, then describes and evaluates the self-characteristics of land-use patch including space, semantics and tense, finally proposes an approach to identify an overall narrow polygon object in map generalization and conducts a related experiment; Conducts description and detection of the contextual characteristics of map object agent by dividing three levels including characteristics of cartographic region, characteristics of group map objects and characteristics of relation between map objects in the scale and distinguishing three characteristics including space, semantics and tense in content, then expresses and evaluates the contextual characteristics of land-use patch including macro level, meso level and micro level, finally calculates the adjacent relation between land-use types in the semantics feature of land-use patch.4. This research proposes a principle what designing agent’s actions (map generalization operators) should fully consider self-and contextual characteristics in multi-agent system for map generalization. This paper reviews the development course of map generalization operators, and points out the shortcomings of the current design of map generalization operators:Put more emphasis on solving spatial characteristics of map object itself, but considers its semantics and temporal characteristics as well as contextual characteristics for less. This research summarizes the content of operators in land-use data generalization, then designs and develops two operators considering self-and contextual characteristics fully in land-use data generalization including subdividing and simplifying the land-use patch.5. This research proposes the basic principles of task decomposing and process controlling, and aiming to settle the complex issue of handling land-use patch in land-use data generalization establishes a decision-making mechanism to judge the importance of the land-use patch while taking into account a variety of criteria, and considers the geometry and semantic indicators integrated into the reasoning process which can be controlled and operated via an automated procedure from judging the importance of the land-use patch to the selection of the scale transformation operators, finally verifies the effects by practical production, the results show that the mechanism can greatly improve the efficiency of map generalization and could keep the areal balance and the feature of full coverage with no overlap and no gap of the land-use data.6. This research builds a thorough hierarchy of quality evaluation for map generalization which considers the multiple constraints under the agent technology, analyzes the reason of causing the map data issues and its content, and proposes the comprehensive control strategies and solutions of the quality issues of map generalization, puts forward the trace tactic for some global quality problems to optimize the process of map generalization thereby contacts the quality evalution to the process compactly in map generalization, and finally conducts a retrospective experiment for area balance in land-use data generalization.7. This research proposes a method for designing and developing software of map generalization based on agent technology. This research divides the map generalization into some key modules under the agent technology, analyzes and discusses multi-agent system for land-use data generalization about selecting of based platform, developping pattern, function partitioning, formulation of technical process, key modules and algorithm designing, then demonstrates and explains the interface of the software, the result and process of handling experimental data, and so on.