230 likes | 397 Views
AAG 2002, Los Angeles. An Intelligent Agent-based Architecture for Internet Mapping and Distributing Geographic Information Services. By Ming-Hsiang Tsou E-mail: mtsou@mail.sdsu.edu Phone: 619-5940205 Fax: 619-5944938. The Department of Geography, San Diego State University.
E N D
AAG 2002, Los Angeles An Intelligent Agent-based Architecture for Internet Mapping and Distributing Geographic Information Services By Ming-Hsiang Tsou E-mail: mtsou@mail.sdsu.edu Phone: 619-5940205 Fax: 619-5944938 The Department of Geography, San Diego State University
The Promise of Internet Mapping Flexible Information Access/Distribution (Spatial Information at your fingertips + Real time data) Information Sharing and Integration (Access multiple Internet map servers at the same time – local data, federal agencies, USGS, EPA, Census etc.) Distributed Mapping/GIS Tools: Web Services (LEGO-like GIS components/mapping tools) 3D, Visualization, Networking Analysis, Hydrological models The Network is Your Mapping Factory
Problems in Internet Mapping Temporary, technology-centered solutions: The lack of an intelligent architecture which can operate in complicated mapping situations and new/unknown environments. Focus on Databases, not on Map Presentation: We need to create a new mechanism for connecting cartographic knowledge with Internet mapping applications.
Information Overload Access hundreds of map layers from different servers How to combine Map-Server-A layers With Map-Server-B layers? Change Symbols? Apply Color Scheme? Create scale threshold? Use Different Projections?
Web Map Users Don’t Have Sufficient Cartographic Knowledge Web Map Users ≠GIS Users • Current Web Maps: • Good Interactivity • Good Flexibility • Poor Quality • Poor Design
Possible Solution: Intelligent Software Agents Apply cartographic principles to web mapping Software Agents (Cartographers pre-defined)+ User defined rules Establish cartographic rules dynamically (Different tasks have different rules and knowledge base) Create distributed cartography knowledge base (CKB) (Access/Distribute different rules and symbols, color schemes, layout.. via the software agent network)
Agent-based Communication GIS Components (Programs) Geodata Objects Metadata Metadata • Software Agents • Info. finders/filters • Interpreters • Decision makers Design User-defined rules Cartography Knowledge Bases Agent collaboration
Information Finder / Filter Generalization Color Scheme Carto-rules Method-1 Rule-2 Landuse-color Method-2 Rule-3 DEM-color ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Rule- 4 Method-400 Zoning-color Color-scheme = DEM Information Finder ------------------------ Information Filter Iso-line rules InterpolationAlgorithm Site-A KEYWORDS Users (User-defined rules) Examples: Search “Color Scheme” for Digital Elevation Model. Search Methods: 1. Message Broadcasting 2. Agent Roaming 3. Create a “Metadata Repository” to improve the search efficiency Site-B Site-C
Information Interpreter [Buffering] in UNIX [Address matching] in Window 2000 Metadata Metadata Metadata becomes the source of knowledge bases Information Interpreter Metadata Metadata Metadata Data-3 (Lat/Lon) Data-1 (UTM) Data-2 (SPCS) • Automatically convert from “UTM” coordinate systems to “SPCS California VI” by accessing the metadata of GIS data objects. • Transform map units from feet to meters. • Transform data from ESRI Shapefiles to AutoDesk SDFs.
The Design of Operational Metadata GeoData Object Map display component Metadata GIS-operation requirements (A, B) (A, B, C, D, E, F) System metadata Integrating Metadata describe how the objects should be represented (color, symbols) and the domain of the object (vector, line, transportation). Other GIS components Self-describing, Self-managing map layers
Decision Maker Agent (decision maker) Actions Events Agent (Information Finder) Agent (Information Interpreter) Agent Collaboration • Event:If a new [polygon] data layer is added into point data layers. • Agent Collaboration: • Info. Finder --> search for cartographic rules • Interpreter --> convert cartographic rules into executable procedures. • Action:Move [point] data layers above the polygon layer.
Software Agents Software Agents User-defined rules (Cartographic rules) User Interface Inference engine Working memory Rules Facts Intelligent software agent Traditional Expert Systems GIS data and components framework GIS Component (Buffering) GIS data object (Road, Colorado) Metadata (Facts) Metadata (Facts) Task #2 (Client Machine–B) (Working memory) Task #1 (Client Machine–A) (Working memory)
Hierarchy of GeoAgents Machine Agent Stationary GeoAgent GeoAgent Component Agent Mobile GeoAgent Geodata Agent Carto Agent UML notations Generalization UML: Unified Modeling Language
Walk-through Example • Sd_pointofinterest • Metadata: • Carto-Type: Point • Symbols: star • Color: red • Size: 7 point • Scale threshold: 1:20,000- 1:10,000 Carto Agents retrieve the operational metadata from data objects and apply it on the map design.
Dynamic Cartographic Design • Add a new landuse layer (metadata: color = blue) • Conflict with current sd_conven layer (same color: blue) • Overlap other information Carto Agents re-arrange the layer sequences and reassign new color scheme for the landuse layer
Above( PointLayer(x), PolyLayer(y) ) Cartography Ontology(Cartography Knowledge Base – CKB ) Statements: All point layers should be above all polygon layers. First-Order Logic: Computer Program: number layout sequences (1:top, 2:second..) polylayer(x).number = i pointlayer(y).number = j If (i < j) then { polylayer(x).number = j pointlayer(y).number = i }
Combining Metadata and Rules Cartographic Rule: If the color of the new polygon layer is the same as one of the existing layers, carto-agents will change the color of new layer to a unique color. Computer Program: Color[AllPolyLayers] = [blue, red, green] Color(NewLayer) = NewLayer.metadata.color While ( Color[AllLayers] contain Color(NewLayer) ) { Color(NewLayer) = Color(Randam) }
Inference by Multiple Knowledge Bases • Multiple Cartographic Knowledge Bases (CKB): • Rule#1: “Landuse” data objects are qualitative. • (from San Diego State University) http://map.sdsu.edu/001.ckb • Rule#2: Color-hue is best visual variable for displaying qualitative area data. • (from UC-Santa Barbara) http://geog.ucsb.edu/hydro.ckb) • Inference: • Rule#1 AND Rule#1 Landuse should use “Color-hue” for area symbol display. • Computer Program (Software Agents): • Landuse.Symbols = ColorScheme(Hue).Attribute(LU)
M M M M M M Implementation: GIService Node GIServices Workstation (a GIS node) : metadata Hardware Profiles: CPU, OS, CRT, printer, scanner GeoData Object Container GIS Component Container M Agent Container Component Agent GeoData Agent Machine Agent Carto Agent
Software Agent Platform • Java (Sun Microsystems): • Java Virtual Machine (VM), Java applets / servlets. • CORBA (OMG): • Common Object Request Broker Architecture, UNIX-based, IIOP (Internet Inter-ORB Protocol), • .NET (Microsoft): • Windows 2000/NT, ActiveX container, COM-based model • XML (W3C):(Extensible Markup Language) • lightweight agent systems, scripting language, • open-ended, metadata- enhanced.
Agent Communication Language (ACL) / Protocol (ACP) • KQML (Knowledge Query and Manipulation Language) (Finin and Weber, 1993) • ACL (Agent Communication Language) specificationFIPA (Foundation for Intelligent Physical Agents) 1997 • IIOP (Internet Inter-ORB Protocol),and CORBA’s • Mobile Agent Facility Specification 1.0 • OMG, (1999) • XML-based scripting language • (Lange, Hill, & Oshima, 2000)
SUMMARY Current Internet Mapping • Poor quality of maps - No cartographic principles • Problems with multiple data/layer presentation • Difficult to apply color schemes / map styles • Unknown situation for mapping new data objects Intelligent Agent Solution • Improve the quality of web maps • Create dynamic cartographic design • Search for appropriate map styles / color schemes. • Establish distributed cartographic knowledge bases.
FUTURE WORK Implementation of Cartography Ontology Convert from “logics and rules” to “computer languages” New Cartographic Principles for New Tasks 3D rules, Layer Transparency, Animation rules, etc. Other A.I. possibility for software agents? Fuzzy logic for scale threshold?, Probabilistic theory for Data uncertainty representation? Neural networks for _______???, PowerPoint Slide is available: http://map.sdsu.edu