170 likes | 333 Views
An Agent Architecture for an Integrated Forest Ecosystem Management Decision Support System. Credits. University of Georgia D. Nute W. D. Potter M. Dass A. Glende H. Uchiyama F. Maier J. Wang USDA Forest Service H. M. Rauscher M. J. Twery, S. Thomasma, P. Knopp.
E N D
An Agent Architecture for an Integrated Forest Ecosystem Management Decision Support System
Credits University of Georgia D. Nute W. D. Potter M. Dass A. Glende H. Uchiyama F. Maier J. Wang USDA Forest Service H. M. Rauscher M. J. Twery, S. Thomasma, P. Knopp
Presentation Overview • Introduction • NED-2 Architecture • NED-2 Ontology and Internal Data Model • NED-2 Blackboard/Database Integration • NED-2 Agents • Future Plans
Introduction NED-2 is an Intelligent Information System for Decision Support for Forest Ecosystems
NED-2: An Intelligent InformationSystem for Decision Support Key design principles: • A single, simple interface • Conversion between data formats without user intervention • Open architecture • Components need no knowledge about other components of the system
NED-2: An Intelligent InformationSystem for Decision Support Features: • Robust ontology • Semi-autonomous agents • Blackboard architecture
NED-2 Architecture External Models Agents Blackboard Prolog Clauses: - Facts - Requests - Plans User Interface Growth & Yield GIS Visualization Access Databases: - User Preferences - Site Data - Inventory Data - Financial Data - Treatment Plans Financial Meta-knowledge Bases Knowledge Bases Inference Engines
NED-2 Ontology and Internal Data Model • NED-2’s design was influenced with practicality as a motive: • Persistent data is stored in data- and knowledge bases • Temporary information is stored as a set of Prolog clauses • Information is stored as snapshots representing the stand at a point in time under a particular treatment regime
NED-2 Blackboard/Database Integration Metaknowledge: Data Structure Blackboard Prolog Clauses PrologQuery Blackboard Handler PrologQuery Data cached in Prolog SQL Agents Access Database Example Query to retrieve information from the blackboard: known(ba([stand(17),plan(`Maximize Timber`),year(2023)], BA)).
NED-2 Agents Overview • NED-2 Interface Agent • NED-2 Treatment Development Agent • NED-2 Simulation Agent
NED-2 Simulation Agent Temporary InputFiles Metaknowledge: Semantics for Simulators Blackboard Access Database Inventory Data Treatment Plans SimulationAgent Simulator Simulated Data Temporary OutputFiles Knowledge Base: Treatment Definitions
NED-2 Agents Overview • NED-2 Interface Agent • NED-2 Treatment Development Agent • NED-2 Simulation Agent • NED-2 Goal Analysis Agent • NED-2 GIS Agent
NED-2 GIS Agent Temporary InputFiles Blackboard Access Database Inventory Data Goal Selections Simulated Data GISAgent GIS Shape File
NED-2 Agents Overview • NED-2 Interface Agent • NED-2 Treatment Development Agent • NED-2 Simulation Agent • NED-2 Goal Analysis Agent • NED-2 GIS Agent • NED-2 Report Writer Agent
NED-2 Report Writer Agent Reports - HTMLFiles Blackboard Report PlanningAgent Access Database Inventory Data: - Goal Selections - Report Selections - Treatment Plans - Simulated Data ReportWriter Plan – Series of Request: - Analysis - Report Writing GoalAnalysis
NED-2 Agents Overview • NED-2 Interface Agent • NED-2 Treatment Development Agent • NED-2 Simulation Agent • NED-2 Goal Analysis Agent • NED-2 GIS Agent • NED-2 Report Writer Agent • NED-2 Planning Agent
Future Plans • Alpha version of NED-2 including functionality described has been completed • Expansions • Improve and extend the goal analysis rules in NED-2 • Understory and ground level simulation models are in development • Development of further regeneration models • Additional overstory growth and yield models • Integration of visualization tools • Integration of a financial analysis model