AGENT SERVICES FRAMEWORK FOR HANDHELD DEVICES SUPERVISOR: DR MUHAMMAD ASLAM Domain: Artificial Intelligence(AI) Muhammad Ali (09-MS-CE-11),
Agenda Background (Agents, Web Services, SOA, FIPA) The Problem Solution(AgentServices) Conclusion QA
Background 1/2 Agent: Many definitions. Agreed definition: Autonomous entities working on behalf of users WebService: Software component available on the web SOA: web service architecture. When a system is distributed into webservices. W3C: standards for web services. FIPA: Standards for agents
Background2/2 Web Services are similar to Agent services. WSDL-No agent equivalent UDDI –DF Access Mechanism: SOAP-FIPA ACL Ontology: is a knowledge represented as Classes, Properties, Domain-Range, Individuals
Agent interaction with its environemnt
The problem JADE Implemented by Communication Vocabularly Partial Ontology support Problem LanguageProtocol Problem: Closed System Implicit Ontology FIPA Specification Problem zone
Most agent implementations are closed systems, and its very difficult for external agents to affectively communicate with them. Problem Statement
Motivation Restaurant Menu on Handheld device Mosque Handheld device on silent
AgentServices Ideal solution: Open System=Agents of different systems can talk and understand each other Interpret agent as Web Service and make it self discoverable and self described Solution: 1. A: Store Ontology within Agent 2. B: Allow transfer of Agent 3. Communication in terms of Individuals and CRM.
Integration of Agent-WebService
AgentService Context
Architecture
Software Hierarchy
Contributions Solves the Domain Knowledge Problem Provide location awareness Specification: Commands and Interaction protocols Library: Android API GUI part: Android client to demonstrate practical aspects
Specification Library Android Client AgentServices Composition
Specification 1/2 Precise set of technical details that specifies that commands and interaction Protocol Ontology Transfer: Command: ASRequestOntology Protocol: Request-Inform Gui Request: Command:AsRequestGui Protocol: Request-Inform
Specification 2/2 Location Properties: Used for DF 1. ASlocationCity: Specifies the city 2. ASlocationCountry: Specifies the country 3. ASlocationAddress: Specifie the address 4. ASlocationPhone: Specifies the phone 5. ASlocationLongitude: Specifies the Longitude 6. ASlocationLatitude: Specifies the Latitude
Library 1/2 Powerful, Easy to Use An API implementing the AS specification Provide classes and wrappers around Jade Service Creation Directory Facilator (DF) AgentServiceRegistration : Class for registration with location DiscoverLocation : A wrapper over the complex Jade service description
Library 2/2 AgentServicesCommands: Provides most common commands AgentServiceMessageHandler : 1. Event based listeners setOnRequestOntologyListener setMessageListener setOnStartActionListener 2. Allows to reply to messages AgentServiceMessageCreator: Wrapper for creating messaages
Client Application 1/5 Android Application to demonstrate the practical aspects of the research. Allows 1. Discovery 2. Search 3. Running of nearby services
Client Application 2/5
Client Application 3/5
App Screenshot 4/5
App Screenshot 5/5
Six easy steps to create an Agent Service Service Creation Steps
Create Protégé Ontology 1/6
Creation based on CRM
Step 2 and 3 Create a new project in eclipse Add the AgentServices.jar library Add the ontology file created in step 1 in source
Place Ontologies in Source
ATM OWL File
Step 4: Register with DF
Step 5 : Implement Essential Commands
Step 6: Implement the Logic Implement the application specific business logic
Conclusion Existing systems are closed, application specific and platforms like JADE don’t support explicit ontology. Allows dynamic creation, registration and discovery. Solves the Domain Knowledge representation problem without using centralized approach or OA Location awareness Future: Streaming, Transfer
Service Discovery and Usage Location Awareness Demo Demonstration
QA