Вы находитесь на странице: 1из 21

Knowledge Based Systems

383 ISM
Dept. of Information Systems

Chapter-6
Agent Based Systems:

Prepared by Zeenat Begum


Topics to be covered
• Agent Based Systems: Introduction.
• Characteristics of agent
• Advantages of agent technology
• Agent communication
• Intelligent Agent
• Multi-agent systems
• Knowledge engineering,
• Agent-based systems - Case Study
What is an Agent?

• An agent is a computational entity that:


– Acts on behalf of other entities in an autonomous fashion
– Performs its actions with some level of proactivity and/or
reactiveness
– Exhibits properties like learning, cooperation, and mobility
to a certain extent
• Software agents (often simply termed agents) are software
systems that freely fit the aforementioned criteria and can
principally be described as inhabiting computers and
networks, assisting users with computer-based tasks.

3
Characteristics of Agents

Characteristic Description

Autonomy Ability to work autonomously without human intervention.


For this purpose, they are supposed to possess the
necessary skills and are enriched with the required
resources.
Cooperation In order to complete the tasks, agents must interact with
users, the environment, and other agents.

Learning Agents should be able to learn from the entities with


which they interact to complete their tasks.

Reactivity Agents perceive their environment and respond in a


timely fashion to changes enforced by the environment.

4
TYPES OF AGENTS
Perceive their environment and responds in timely fashion to changes enforced
by the environment.
1. Proactive: Agents that can exhibit goal oriented behavior by taking initiative
are called as proactive agents.
2. Reactive: Agents that act in response to their environment when provoked
by events are called reactive agents.

Property of an Agent
1.Veracity: An agent must not knowingly communicate false information.

2.Benevolence: An agent perform its duty without any confusion.

3.Rationality: Helps an agent to act in order to achieve the goals and they will
not act in such a way as to prevent their goals from being achieved.

4.Social ability: Ability to interface with other agents via a communication


language.
Architecture of an Agent

Action

Autonomy

Cooperation Environment
Sensors
Learning
Effectors
Reactivity

Perception

Figure 6.1: Architecture of an agent


Advantages of a Agent

• They can be used to solve large, complex problems.

• They allow for the interconnection and interoperation of


multiple existing legacy systems.

• They provide solutions to problems where information


resources, expertise, and the problem itself are widely
distributed.

• They enhance modularity, speed, reliability, flexibility, and


reusability in problem solving.

• They lead to research into other issues—for example,


understanding interactions among human societies.
Agent Communication Languages
An agent Communication language is of three
types
1. Inner language: Knowledge interchange
Format(KIF)
2. Outer language: Knowledge Query and
manipulation language
3. Common Vocabularies(i.e ontologies)
1.Knowledge Interchange Format(KIF):
It is generic formal representation of the expression of the
internal knowledge base of an agent. An executable agent can
apply a specific formal knowledge base of an agent
Representation of the knowledge, translate it into KIF and then
communicate it to others.
2.Knowledge Query and Manipulation
Language(KQML):
• Agent communication language based on speech act theory.
• It is a general purpose high level message oriented
communication language.
• It facilitates interoperability and communication between
software agents.
• It is a protocol for information exchange independent of
content syntax and ontology.
There are three layers in KQML message:
• Content layer: Actual content in the programs own
representation language .
• Communication layer: Encodes a set of feature to the
message that describes the lower level communication
parameters.
• Message layer: Identifies the network protocol with
which to deliver the message and supplies a speech act
that the sender attaches to the content.

• Common Vocabularies(i.e. ontologies): These are


the controlled vocabularies providing the basic
terminology necessary for representation of a
domain of discourse. They are the fundamental
elements of knowledge media.
Agent Communication Languages

Knowledge Query and Manipulation Language (KQML) block


(a query about the price of a share)

(ask-one
:content “price (Infosys, [?price])”
:receiver stock-server
:language LPROLOG
:ontology NYSE-TICKS)

(ask-all
:content "price(Infosys, [?price, ?time])"
:receiver stock-server
:language standard prolog
:ontology NYSE-TICKS)
Figure 6.2: KQML block for agent communication

11
Intelligent agent
• It is software system that can send information to and receive it
from other agents using appropriate protocols.
• It is defines as agents capable of flexible autonomous action to
meet their designs objectives, create action plans, process the
information received and perform reasoning.
Features of Intelligent Agent:
 Events and tasks list: Tasks that the agent can perform within the
environment.
• Controller: Controlling and communication activities.
• Goals: Objectives to be accomplished.
• Knowledge base: Consists of facts, beliefs, rules etc..

12
Intelligent Agents

Controller

Knowledge Base

Executable Goals and


Tasks List Inference Objectives

Interface

I/O Queue

Figure 6.7: Structure of an intelligent agent

13
Multiagent Systems

• A multiagent system is composed of several intelligent agents working


together toward a goal or completion of a task.
• It is a loosely coupled network of problem-solving entities that work
together to find answers to problems that are beyond the capacity of any
individual problem-solving entity.
• This system is called for when complex problems require the services of
multiple agents with diverse capabilities and needs.
• Besides multiple agents, a multiagent system (MAS) does the following:
– Provides an environment for the agents
– Sets the relationships between the entities
– Provides a platform for a set of operations that can be performed by the
agents

14
Architecture of a Generic Multiagent System
Repository of data and knowledge

Knowledge base

Optional distributed databases

Middle agent services and agent meeting place

Domain services
Domain agent 1 Domain agent 2 Domain agent N

Control and presentation services


Master agent Interface agent Information agent

Optional link

Figure 6.8: Layered architecture of generic multiagent system


15
Control and Presentation layer:
• Master agent interacts with the other the agents of the
systems a swell as with the repository layer(knowledge bas
and database).
• Interface agent is required with the master agent to facilitate
users interaction with the system.
• This layer can be used to manage load profile and
Documents ,storing, users profile
• Middle layer serves as a meeting place for agents, facilitating
communication between them

16
Advantages of Multiagent systems
• Able to solve large problems.
• Different systems to work together in an
interconnected fashion.
• Provide efficient solution where information is
distributed among different places.
• Software reusability: More flexible in adapting
different agent capabilities for problem
solving.
Interface Agents(IA):
They work a personnel assistant, helping users to interact
with the system. They provide a user friendly environment to
work with a highly technical application

Examples: IA can be used in application as


Personnel assistant
Tutorial Systems
Application
Financial decision Systems User
User’s
agent Optional
link
User

Other
Agent
Figure 6.3: An example of an interface agent

18
Case Study
Architecture of a TravelPlan Multiagent System

Optional Link

WebBot

Problem World
PlannerAgent WebBot
UserAgent
Wide
Solution
Web
WebBot
Problem
UserAgent
PlannerAgent
Solution
WebBot

Figure 6.9: TravelPlan general architecture

19
Communication Protocol in TravelPlan

UserAgent PlannerAgent WebBot

REQUEST.ORDER
(Prob,PA) REQUEST.TO.DO.ORDER(Prob)

ACCEPT(UA)
REQUEST.ORDER
(Prob,UA)
REQUEST.PROPOSE(Subprob,WB) REQUEST.TO.DO.
PROPOSE(Subprob,WWW)

Problem REJECT(PA)

QUERY.ORDER(Subprob,WB)
User
ACCEPT(PA)

Solution
INFORM(Solution, Subprob)

INFORM(Solution, Prob)

Figure 6.10: Example of the communication protocol in TravelPlan

20
Agents in Travel Plan System
• User Agent: It pays attention to users queries and
shows the solution. It analyzes the problem and
obtain an abstract representation. It give solution
to the problem to Planner Agent.
• Planner Agent: are designed to work with User
Agent and determine a set of possible solution to
a problem.
• WebBot: these agents fill in the details
(requested by Planner Agent) obtaining the
required information from the Internet.