Scribd
Upload
69 views17 pages

Intelligent Information Retrieval: Submitted By: Deeksha Parsediya 13/pit/003 (ISR)

The document describes an intelligent agent system for information retrieval. It discusses designing a universal agent model structure that can be used to quickly design and implement specific agents for multi-agent systems. It also describes the roles and design of control agents, human-computer interaction agents, and various types of retrieval agents. Adaptation mechanisms are discussed at both the individual agent and organizational levels to allow agents to effectively respond to unexpected events or dynamic environments.

Uploaded by

dixhap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
69 views17 pages

Intelligent Information Retrieval: Submitted By: Deeksha Parsediya 13/pit/003 (ISR)

The document describes an intelligent agent system for information retrieval. It discusses designing a universal agent model structure that can be used to quickly design and implement specific agents for multi-agent systems. It also describes the roles and design of control agents, human-computer interaction agents, and various types of retrieval agents. Adaptation mechanisms are discussed at both the individual agent and organizational levels to allow agents to effectively respond to unexpected events or dynamic environments.

Uploaded by

dixhap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 17

INTELLIGENT

INFORMATION RETRIEVAL

Submitted by:
Deeksha Parsediya
13/pit/003(ISR)

Deigning Agent system for intelligent information retrieval


The Agent General Structure Design
Establish the Universal Agent model structure for design Multi-Agent

Systems.
All the specific Agent of Multi-Agent Systems Agent can be quickly
designed and implemented on the base of the Universal Agent
structure model
The structure of the model not only can sense changes in the external
environment to react accordingly, based on changes in their own state
of mind to think smart manifestation.

Agent Design, Human-computer


Interaction.

Control Agent Design


Control Agent bear the role of the hub of the entire Multi-Agent Systems. It is

the key to run the entire system effectively, to work together with Multi-Agent
effectively. The Agent interaction must organized as an intermediate control
by controlling the Agent.
Firstly, Control Agent receive the request message from the Human-

Computer Interaction Agent.


Secondly, word processing Agent to complete the processing of the
segmentation will notify the control Agent.
Finally, whenever received confirmation of completion messages from a
specific retrieval Agent ,it is necessary to check receipt of all retrieved Agent
If not, continue to wait for the listener.

Listener Implementation Based on Socket

public class Monitor Thread extends Thread


{
//Other variables
public final int socket port;//socket Port
number
//Other methods
public void run()
{
Server Socket ss=null; Socket s=null;
try {
ss = new ServerSocket(socket_port);//make a
new ServerSocket
while((s=ss.accept())!=null){
Message mess=null;//information
Input Stream is=s.getInputStream();
ObjectInputStream ois=new Object Input
Stream(is);
try {
mess=(Message)ois.readObject();
} catch (ClassNotFoundException e) {
e.printStackTrace();
}
//process information
ois.close();//close the flue
s.close();//close Socket objects
}
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
finally{
ss.close();
Fig .5 Listener Implementation Based on Socket

Intelligent Adaptive Information Agents.


Adaptation is behavior of an agent in response to

unexpected (i.e., low probability) events or dynamic


environments.
several types of adaptation supported by the
individual agent
Organizational Adaptation
Planning Adaptation
Scheduling Adaptation
Execution Adaptation

Agent Control Component

The control process for information agents includes steps for communicating

agent goals or objectives, planning to achieve local or non-local objectives,


scheduling the actions within these plans, and actually carrying out these
actions.

Organizational Adaptation
In multi-agent information systems, one of the most important sources

of uncertainty revolves around what information is available from


whom (and at what cost).
In this developed a standard basic advertising behavior that allows
agents to encapsulate a model of their capabilities and send it to a
matchmaker information agent. Such a matchmaker agent can then
be used by a multi-agent system to form several different
organizational structures.
Uncoordinated Team: agents first query the matchmaker as to who
might answer the query, and then choose an agent randomly for the
target query.
Federations: agents give up individual autonomy over choosing who
they will do business with to a locally centralized facilitator (an
extension of the matchmaker concept) that brokers requests.
Centralization of message traffic potentially allows greater load
balancing and the provision of automatic translation and mediation
services.

Planning Adaptation
The planner portion of our agent architecture consists of a new

hierarchical task network based planner using a plan formalism that


admits sophisticated control structures such as looping and periodic
tasks.
It has features derived from earlier classical planning work, as well as task
structure representations. The focus of planning in system is on
explicating the basic information flow relationships between tasks, and
other relationships that affect control-flow decisions.
At any time, any agent in the system might be unavailable or might go offline.
planners task reductions handle these situations so that such failures are
dealt with smoothly.
If alternate agents are available, they will be contacted and the
subproblem restarted.If no alternate agent is available, the task will have
to wait. In the future, such failures will signal the planner for an
opportunity to replan.

Scheduling Adaptation
The agent scheduling process in general takes as input the

agent's current set of task structures T , in particular, the set of all


basic actions, and decides which basic action, if any, is to be
executed next.
Constraints
on the scheduler include:
No action can be intended unless it is enabled.
Periodic actions must be executed at least once during their
period
Actions must begin execution before their deadline.
Actions that miss either their period or deadline are considered to
have failed. the scheduler must report all failed actions.
Sophisticated schedulers will report
such failures before they occur by reasoning about action
Durations.
.

Execution Adaptation
In this looking at load-balancing/rebalancing behaviors such as

agent cloning.
Cloning: Cloning is one of an information agents possible responses to
overloaded conditions. When an information agent recognizes
via self-reflection that it is becoming overloaded, it can remove
itself from actively pursuing new queries and create a new
information agent that is a clone of itself.
it uses a simple model of how its ability to meet new deadlines
is related to the characteristics of its current queries and other
tasks. It compares this model to a hypothetical situation that
describes the effect of adding a new agent.

Here we focus on database monitoring queries only.


Each monitoring goal is met by a task that consists

of three activities; run-query, check-triggers, and


send-results.
If we let E be the average time to process an
external query, then with n queries of average period
p, we can predict an idle percentage of:
I% = (p En) / p

Execution Adaptation: Experimental Results


In it measurement of the baseline performance of an information agent.

A graph of the average percentage idle time and average percentage of


actions with deadlines that missed them for various loads (left Y axis).
Superimposed on the graph, and keyed to the right axis, are the average
number of seconds by which a missed deadline is missed.

You might also like