Fuzzy Mobile Agents for Distributed e-Shopping Data Mining Presented by Lin Lu.

Fuzzy Mobile Agents for Fuzzy Mobile Agents for

Distributed e-Shopping Data MiningDistributed e-Shopping Data Mining

Presented byLin Lu

AcknowledgementAcknowledgement

First of all, thanks to Dr. Zhang for guidance, encouragement and patience throughout the length of the project

Thanks also go to my committee member, Dr. Sunderraman, for his continuous support over the time of my stay at GSU

OverviewOverview

Introduction Architecture of KAARIBOGA

Mobile Agents Design Issues of FMADeSDM Implementations of FMADeSDM Concluding Remarks Demo

IntroductionIntroduction Background and purpose• Explosive growth of World Wide Web (WWW)

makes retrieving information of interest dramatically more challenging

• Currently-used smart commercial search engines always fall short in providing prompt and efficient results

• Mobile agent paradigm has been recently developed, with high demands in e-commerce applications

• Demands for Intelligent mobile agent

IntroductionIntroduction What is mobile agent? A mobile agent is an autonomous program that

can migrate through a heterogeneous network searching for and interacting with services on user's behalf.

What is fuzzy logic? Fuzzy logic is a superset of conventional

(Boolean) logic that has been extended to handle the concept of partial truth -- truth values between "completely true" and "completely false".

Introduction (cont.)Introduction (cont.) Why fuzzy logic?

• Fuzzy logic uses soft linguistic variables to represent the range of numerical values and allow these linguistic values to overlap.

• Fuzzy logic can be used to deal with uncertain information to come up with decisions, which is ideal for solving real-world problems.

Algorithms used in the project• Fuzzy-user-preference-based ranking

algorithm• Distributed data mining algorithm and

centralized data-mining algorithm

Architecture of KAARIBOGA Architecture of KAARIBOGA Mobile AgentsMobile Agents

Life-cycle Model – Creation – Start – Destroy – Dispatch – Arrival – Sleep – Awake

– Message handling

Architecture of KAARIBOGA Architecture of KAARIBOGA Mobile Agents (cont.)Mobile Agents (cont.)

Navigation Model

Send Kaariboga Message

Kaariboga Base Kaariboga Base

Transfer agent between Kaariboga bases

Pack agent into message

Unpack agent

Architecture of KAARIBOGA Architecture of KAARIBOGA Mobile Agents (cont.)Mobile Agents (cont.)

Communication Model

Message exchange between agents and/or bases

Kaariboga Base Kaariboga Base

Kaariboga Domain

Architecture of KAARIBOGA Architecture of KAARIBOGA Mobile Agents (cont.)Mobile Agents (cont.)

Architecture of Kaariboga System

Architecture of Kaariboga system

Kaariboga Base Kaariboga Base

Design Issues of Design Issues of FMADeSDMFMADeSDM Fuzzy Ranking

1.0

0

low medium high

Min. (Min.+Max.)/2 Max. Price

Fuzzy linguistic values for price

1.0

0

short medium long

Min. (Min.+Max.)/2 Max. Distance

Fuzzy linguistic values for distance

1.0

0

very low medium very high

low high

0 0.25 0.5 0.75 1.0 Rank0.083 0.917Fuzzy linguistic values for rank

Price Distance

Low Medium

Short Very High High Medium

Medium High Medium Low

Long Medium Low Very Low

High

Fuzzy rule base for Price, Distance and Rank

Design Issues of Design Issues of FMADeSDMFMADeSDM(cont.)(cont.) Fuzzy Ranking Example

1.0

0

low medium high

180 200 220 Price

Fuzzifications for price = $185

0.75

0.25

185

1.0

0

short medium long

5 15 25 Distance

Fuzzifications for distance = 17miles

0.8

0.2

17

PiDi

iPiDiRank =

(0.75*0.75*0.8+0.5*0.75*0.2+0.5*0.25*0.8+0.25*0.25*0.2)

(0.75*0.8+0.75*0.2+0.25*0.8+0.25*0.2) = = 0.64

Fuzzy rule for Price = $185, Distance = 17mile

Price Distance

Low MediumMedium High Medium

Long Medium Low

Shopping Searching Agents• Search Agent 1

storeuser

Search Agent

dispatch

Local Agent

generate

goSearch Agent

result

Local File

search messagewith result

Search Agent

time out go

Scenario for search agent 1

Design Issues of Design Issues of FMADeSDMFMADeSDM(cont.)(cont.)

Search Agent

dispatch

user 1store

2store

Shopping Searching Agents• Search Agent 2

Scenario for search agent 2

go

Local Agent

generate result

Local File

search messagewith result

go

result

messagewith result

Fuzzy Ranking Display

go

Search Agent

time outcounter=1

Search Agent

time outcounter=2

go Search Agent

search

Local File

go

Search Agent

Design Issues of Design Issues of FMADeSDMFMADeSDM(cont.)(cont.)

Shopping Searching Agents• Search Agent 3

Scenario for search agent 3

Search Agent

dispatch

Local Agent

generate

go

Local File

resultsearch

Search Agent

messagewith rank

time out go

resultsearch

message with rank go

go

Search Agent

counter=1time outcounter=2

go

user Search Agent

Local File

Search Agent

1store

2store Fuzzy

Ranking

Fuzzy Ranking

Design Issues of Design Issues of FMADeSDMFMADeSDM(cont.)(cont.)

Update FuzzyValue

Implementation of Implementation of FMADeSDMFMADeSDM Fuzzy Ranking

Personalized fuzzy ranking criteria

Implementation of Implementation of FMADeSDMFMADeSDM(cont.)(cont.) Search Agent 1

Interface for dispatching search agent 1

Search result of search agent 1 (b)

Search result of search agent 1 (a)

Message on visited store server

Search Agent 2 & 3

Interface for dispatching search agent

Search result of search agent

Implementation of Implementation of FMADeSDMFMADeSDM(cont.)(cont.)

Concluding RemarksConcluding Remarks

Kaariboga Mobile Agents system is introduced and studied

Fuzzy Mobile Agents for Distributed e-Shopping Data Mining System is developed

Implemented three kinds of search mobile agents

Proposed a simple scenario to monitor the aliveness of each search agent

Concluding Remarks (cont.)Concluding Remarks (cont.)

Fuzzy-user-preference-based ranking algorithm is used

Dynamically updated fuzzy values are employed in distributed data mining algorithm

Ideas proposed in FMADeSDM can be extended to similar applications beyond the e-commerce application

DemoDemo