INFORMATION AND KNOWLEDGE MANAGEMENT

coilcruelΔιαχείριση

6 Νοε 2013 (πριν από 3 χρόνια και 5 μήνες)

47 εμφανίσεις

INFORMATION AND KNOWLEDGE MANAGEMENT (SBEM)

Lecture: block lecture with oral exam at the end of the block

Language of instruction: English

Course rationale

Economics and management include core studies in complex systems involving all areas of system sciences mainly complexities
involving information flow and information structuring into knowledge and its representation. Tools needed in such studies
include mathematics with soft modelling, decision support, simulation and knowledge representation. The course outlined in
this document is designed as an essential part of this educational process.

Course description:
The lecture will cover general field of information and knowledge management.

Typical lecture topics include the following:




Modern information systems




Decision support systems




Information modelling




Information structuring into knowledge




Knowledge representation




Knowledge management




Soft modelling and Artificial Intelligence techniques




Decisional DNA




E
-
Community

Content and mode of delivery:
LECTURE AND DISCUSSIONS

Grading policy: Oral exam


100%




Text:
Reading and research related to the lecture is based on material available in the library or on the Internet. The lecture wil
l b
e
based in some parts on the books:



E Szczerbicki “
Information Management: Modelling, Analysis, and Simulation Perspective
”, GTN Gdansk, 2004.


Szczerbicki, E; Nguyen, N “
Smart Information and Knowledge Management: Advances, Challenges, and Critical
Issues”,
Springer Berlin, 2010.



N Thanh, Szczerbicki, E “
Intelligent Systems for Knowledge Management”,

Springer Berlin, 2009.




Contact with course coordinator:


Prof E Szczerbicki:
Edward.Szczerbicki@zie.pg.gda.pl

Gmach B, 820 ph 21
-
95


Information and knowledge management links are easy to find using any of the popular search engines. Typical links
are also listed on the page
http://www.zie.pg.gda.pl/zwi/
. You can also find lecture notes there.

INFORMATION AND
KNOWLEDGE MANAGEMENT
IN COMPLEX SYSTEMS

Lecture notes:


http://www.zie.pg.gda.pl/zwi/


-
Nine parts of presentations (over 400 slides)

-
Two recent conference papers on case
studies




Ksiazka jest do nabycia w GTN, Gdansk.


Zamowienia mozna przesylac do

Biura Towarzystwa e
-
mailem
gtn@3net.pl

, faksem
305
-
81
-
31.



Ksiazka jest wysylana poczta z faktura VAT.


Sprzedaz prowadzi tez Ksiegarnia Naukowa przy ul.
Lagiewniki 56 w Gdansku









Intelligent Systems for Knowledge Management

Series:
Studies in Computational Intelligence

,
Springer Berlin.


Nguyen, Ngoc Thanh; Szczerbicki, Edward

2009, XII, 332 p. 147 illus., Hardcover

ISBN: 978
-
3
-
642
-
04169
-
3




Springer

Customer Service Center GmbH

Haberstrasse 7

69126 Heidelberg

Germany


Call:
+ 49 (0) 6221
-
345
-
4301


Fax:
+49 (0) 6221
-
345
-
4229


Web:
springer.com


Email:
orders
-
hd
-
individuals@springer.com

Smart Information and
Knowledge Management:
Advances, Challenges, and
Critical Issues

Series:
Studies in
Computational Intelligence

,
Vol.

260

Szczerbicki, Edward;
Nguyen, Ngoc Thanh (Eds.)


2010, X, 340 p., Hardcover

ISBN: 978
-
3
-
642
-
04583
-
7




SOCIAL CHARACTERISTICS OF
THE AGRICULTURAL, INDUSTRIAL
AND INFORMATION SOCIETIES

(
Castells 2001, 2003
)

AGRICULTURAL

SOCIETY

INDUSTRIAL
SOCIETY

INFORMATION

SOCIETY

TECHNOLOGICAL
CONTEXT

FARM

FACTORY

ELECTRONIC
MEDIA

SOCIAL SETTING

NATURAL

INDUSTRIAL

ARTIFICIAL

S
PACE/TIME
RELATIONSHIPS

PHYSICAL

(NATURAL)

PHYSICAL

(MAN
-
MADE)

PSYCHOLOGICAL

PLANNING

SEASONAL

ECONOMIC CYCLES

PROCESS
ORIENTED

(MULTIPLE
HORIZONS)

BASIS FOR SECURITY

COMMUNITY

COOPERATION

INSURANCE

NETWORKS

BASIS FOR HOPE

RELIGIOUS FAITH

SCIENCE/

TECHNOLOGY

HUMAN

INTERACTION

MANAGEMENT OF
INFORMATION…

Presentation outline:

i.
significance and aims

ii.
methods and techniques

iii.
results so far

iv.
implementations

SIGNIFICANCE AND AIMS


If

all

the

N

elements

of

a

system

are

required

to

communicate,

the

amount

of

information

transfer

is

likely

to

become

unmanageable
.



The

above

has

been

the

reason

why

systems

that

are

divided

into

smaller

subsystems

(called

atomized

or

multi
-
agent

or

multi
-
component

systems)

are

recently

gaining

considerable

attention
.



In

atomized

approach

efficiency

of

components

depends

on

quality

and

quantity

of

information

flow

(Jain

2001
)


Our

research

is

primarily

concerned

with

capture

of

knowledge

useful

in

structuring

and

evaluation

of

such

an

information

flow
.

AUTONOMOUS
AGENTS/SYSTEMS


AUTONOMOUS

AGENTS

CONSIST

OF

GROUPS

OF

PEOPLE,

MACHINES,

ROBOTS,

AND/OR

GUIDED

VEHICLES

TIED

BY

THE

FLOW

OF

INFORMATION

BETWEEN

AN

AGENT

AND

ITS

EXTERNAL

ENVIRONMENT

AS

WELL

AS

WITHIN

AN

AGENT



AUTONOMOUS

AGENTS

CAN

STILL

BE

INTERRELATED

AND

EMBEDDED

IN

LARGER

SYSTEMS,

AS

AUTONOMY

AND

INDEPENDENCE

ARE

NOT

EQUIVALENT

CONCEPTS
.


(E
.

Szczerbicki,

IEEE

Transactions

on

Systems,

Man,

and

Cybernetics,

Vol
.

23
,

p
.

1302
)

SOLVING COMPLEX
PROBLEMS

COMPLEX

SYSTEM

REPRESENTATION

DECOMPOSITION

INTEGRATED

SOLUTION

INTEGRATION

DECOMPOSITION…..

TASK DECOMPOSITION
GLOBAL TASK
T1
T2
T3
T4
T5
T6
T7
T8
T9
T10
SPECIFY OR
SELECT
A FUNCTION
SPECIFY OR
SELECT
A FUNCTION
SPECIFY OR
SELECT
A FUNCTION
SPECIFY OR
SELECT
A FUNCTION
SPECIFY OR
SELECT
A FUNCTION
SPECIFY OR
SELECT
A FUNCTION

DECOMPOSITION….

FUNCTIONAL DECOMPOSITION SPACE
FUNCTION
F1
F2
F3
F4
F5
F6
F7
F8
F9
F10
F11
F12
F13
F14
F15
F16
F17
F18
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT
SPECIFY
OR
DEVELOP
AN AGENT

AND/OR CLAUSES

AND/OR CLAUSE REPRESENTATION
GRAPHICAL
LITERAL
A
B
A OR B
A
B
A AND B
A
B
C
A AND B OR C
A
B
C
A AND B OR B AND C
A
B
C
A AND B AND C

EXAMPLE:

AND/OR NOTATION

For example if task T1 is specified as:



T1: Manufacture a hole


the corresponding functional sets

may be listed as follows:



S
1
= {(F1 AND F2) OR F3}



F1: Centre drill



F2: Trepan



F3: Gundrill


S
2
= {(F1 AND F4) OR F4}



F4: Twist drill

SOLVING COMPLEX
PROBLEMS

COMPLEX

SYSTEM

REPRESENTATION

DECOMPOSITION

INTEGRATED

SOLUTION

INTEGRATION

QUESTIONS…….


How

to

structure

an

exchange

of

information

between

a

system

and

its

uncertain,

dynamic

and

imprecise

environment?



What

is

better,

complete

information

but

heavily

delayed,

or

incomplete

information

less

delayed?


losses

amount of information

min

losses

informational

balance

incompleteness


delay

total losses

INFORMATIONAL BALANCE



Informa
tion: realization of variables describing the state

of external environment in epoch T







X1(T) X2(T) .... XN(T)


Distribution of information: sensoring and
communication between the agent elements


Information structure C: the

resulting flow of information

within the agent


Agent decision making process:

choosing the optimal action based


on the agent knowledge
describing given decision situation


Information flow

evaluation: VC

Recommendations for

sensoring and

communication


Characteristics of

external environment:




Characteristics of

internal environment:



EVALUATION OF
INFORMATION FLOW


A
OPT
ACTION
A
ENERGY
E
E = E
1
MIN
ENERGY
E
INFORMATION
C
E
C
1
1

ENERGY, INFORMATION AND
ACTION

X1

X2

X3

X4

1 0 0 0

0 1 0 0

0 0 1 1

0 0 1 1

INFORMATION STRUCTURE




VC=min E[f(a, X(t))| C0]
-
minE[f(a, X(t))|C]






)
(
1
1
)
0
(
1
)
1
(
)
(
2
0
2
2
1
2
1
0
j
k
p
i
i
j
t
k
p
i
i
t
X
k
p
i
i
t
X
t
X
i
j
t
i
i
t
j
i
t
i
i
i
t
i
i





















































THE VALUE OF
INFORMATION STRUCTURE C

SOFT VS HARD MODELLING

SOFT MODELLING

PHYSICAL MODELLING

THEORY RICHNESS

DATA RICHNESS

EXPERT SYSTEMS

NEURAL
NETWORKS


correlation (R) dynamics (T) interaction (I) del
ay (D) process (W)

decisions concerning the
flow

of information

observation

exchange of information

characteristics of an agent

NEURAL NETWORKS

no

value

description

observation

exchange




r
=0.95


strong relationship

between variables

describing external

environment




t
=0

external environment

is static



1

q
=0.01

there is no

interaction in

internal environment

yes

no


d
=0

information is not

dela
yed




w
=0

process is

independent







r
=0.2


weak relationship

between variables

describing external

environment



2

t
=0

external environment

is static




q
=0.90

there is interaction

in internal environment

yes

yes


d
=0

information is

not delayed




w
=0

process is independent





THE USE OF THE TRAINED
NETWORK



(Quinlan 1990):

Step 1.

If all the objects in S belong to the same class,
for example C
i
, the decision tree for S consists of
a leaf labelled with this class.

Step 2.

Otherwise, let T be some test with possible
outcomes
O
1
, O
2
, ..., O
n
. Each object in S h
as one
outcome for T so the test partitions S into
subsets S
1
, S
2
, ... S
n
where each object in S
i
has
outcome
O
i
for T. T becomes the root of the
decision tree and for each outcome
O
i
we build a
subsidiary decision tree by invoking the same
procedure recur
sively on the set S
i
.


Training set for agent functioning

external

environment


internal

environment

type of

dynamics

correlation

delay of

information

decision

static


independent_actions

0

0

0

no

static


independent_actions

0

0.7

1

no

dynamic


dependen
t_actions

1.5

-
0.5

1

yes

static


dependent_actions

0

0

0

yes

static


independent_actions

0

1

2

no

static


dependent_actions

0

0.5

2

yes

static


dependent_actions

0

-
1

3

no

static


independent_actions

0

-
1

0

no

static


dependent_actions

0

0.9

1

yes

s
tatic


dependent_actions

0

1

1

no


DECISION TREE CLASSIFIERS



internal_environment

independent_action
s

dependent_actions

do_not_exchange_information


-
1 < correlation < 1

true

false

do_not_exchange_information

exchange_information

DECISION TREE: TESTING ON
INTERNAL ENVIRONMENT





















SDG model of informational balance of an agent


w

d

a

s

q

L2

L1

+

+

+

+

+

+

+

+

+

_

LV

+

+

SIGNED DIRECTED GRAPHS
(SDG)


SDG MODEL OF AN
INFORMATIONAL BALANCE


FINAL SIMPLIFICATION









SDG Rule 1:

IF [d=+] .and. p[dLV]

THEN it is a possible solution pattern for a positive change in d

SDG Rule 2:

IF [a=+] .and. n[aLV]

THEN it is a possible solution pattern for a positive change in a

SDG Rule 3:

IF[w=+] .and. p[wLV] .and. p
[wd] .and. p[dLV]

THEN it is a possible solution pattern for a positive change in w


_

w

d

a

LV

+

+

+

SAMPLE OF PRODUCTION
RULES

RULE

12

IF

an

external

environment

of

an

autonomous

agent

is

static,

AND


there

is

an

interaction

in

the

internal

environment,

AND

the

relationship

between

variables

describing

the

external


environment

is

of

statistical

character,

THEN

information

structure

should

include

observation

(sensoring)


and

communication
.

RULE

13

IF

an

external

environment

of

an

autonomous

agent

is

static,

AND

the

relationship

between

variables

describing

the

external


environment

is

given

by

function

dependence,

THEN

communication

between

agent

elements

does

not

affect

the

value


of

information

structure
;

information

flow

should

be

restricted


to

observation

(sensoring)
.

SOLVING COMPLEX
PROBLEMS

COMPLEX

SYSTEM

REPRESENTATION

DECOMPOSITION

INTEGRATED

SOLUTION

INTEGRATION

SY STEM 7
SUB SY STEM S
S6
SUB SY STEM S
S5
SUB SY STEM S
S3
AGENTS
S1
S4
S2
A5
A3
A1
A4
A7
A8
A2
A6
AGENTS
AGENTS
AGENTS

FIVE LEVEL HIERARCHICAL
TREE OF THE OVERALL
INTEGRATED SYSTEM

INTEGRATION……

Reference