business rules for by benjamin grosof ibm copyrights reserved€¦ ·  · 1999-09-22commonrules by...

52
9/22/99 by Benjamin Grosof IBM copyrights reserved Business Rules for E-Commerce: Interoperability and Conflict Handling Dr. Benjamin Grosof IBM T.J. Watson Research Center Hawthorne, NY, USA [email protected] http://www.research.ibm.com/people/g/grosof

Upload: truongtuyen

Post on 13-Apr-2018

215 views

Category:

Documents


2 download

TRANSCRIPT

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Bu

sin

ess

Ru

les

for E-C

om

me

rce

:In

tero

pe

rab

ility

an

d C

on

flict

Ha

nd

ling

Dr.

Ben

jam

in G

roso

fIB

M T

.J. W

atso

n R

esea

rch

Cen

ter

Haw

thor

ne, N

Y, U

SA

gros

of@

us.ib

m.c

om h

ttp://

ww

w.r

esea

rch.ib

m.c

om/p

eopl

e/g/g

roso

f

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

edOve

rvie

w•

Intr

o &

Bac

kgro

und

–O

vera

ll m

issi

on, t

echn

ical

app

roac

h, in

dust

ry tr

ends

–E

EC

OM

S p

roje

ct (

NIS

T A

TP

) on

inte

r-en

terp

rise

supp

ly c

hain

inte

grat

ion

•C

ore

Tec

hnol

ogy:

Com

mon

Rul

es pr

otot

ype

base

d on

logi

c pr

ogra

m K

R

–In

nova

tive

conf

lict h

andl

ing,

pro

cedu

ral a

ttach

men

ts; s

ampl

e en

gine

–X

ML

Inte

rling

ua b

etw

een

hete

roge

neou

s ru

le s

yste

ms;

sta

ndar

ds p

lay

•A

pplic

atio

ns:

repr

esen

t bus

ines

s pr

oces

ses/

wor

kflo

w, p

olic

ies,

pro

duct

s

–C

ontr

acts

/agr

eem

ents

, neg

otia

tion

esp.

B2B

, sec

urity

aut

horiz

atio

n

–E

EC

OM

S s

uppl

y ch

ain

scen

ario

s, N

et.C

omm

erce

-typ

e B

2C s

cena

rio

•D

irect

ions

–E

xter

nal a

lpha

rel

ease

in a

bout

7/9

9 of

cor

e te

chno

logy

–E

xplo

re m

ore

appl

icat

ions

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Vis

ion

ove

rall

•V

isio

n:

Rul

es a

s an

impo

rtan

t asp

ect o

f com

ing

wor

ld o

f Int

erne

t e-

busi

ness

: r

ule-

base

d bu

sine

ss p

roce

sses

for

both

B2B

and

B2C

.

–re

pres

ent b

uyer

’s r

eque

sts,

inte

rest

s, b

ids

–re

pres

ent s

elle

r’s o

fferin

gs o

f pro

duct

s &

ser

vice

s, c

apab

ilitie

s,bi

ds; m

ap o

fferin

gs fr

om m

ultip

le s

uppl

iers

to c

omm

on c

atal

og.

–re

pres

ent b

usin

ess

proc

esse

s, e

.g.,

sale

s he

lp, c

usto

mer

hel

p,pr

ocur

emen

t, au

thor

izat

ion,

bro

kerin

g, w

orkf

low

.

–au

tom

atic

exe

cutio

n; m

atch

mak

ing

of b

uyer

s w

ith s

elle

rs

–hi

gh le

vel o

f con

cept

ual a

bstr

actio

n, e

asie

r un

ders

tand

ing

and

spec

ifica

tion

by n

on-p

rogr

amm

ers

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ru

les:

F

un

da

me

nta

l Te

chn

ica

l Ap

pro

ach

•A

im to

ena

ble:

exc

hang

e &

upd

ate

busi

ness

rul

es,

dyna

mic

ally

.

•C

onte

xt:

key

appl

icat

ion

logi

c is

rep

rese

nted

via

rul

es, i

nm

any

syst

ems.

E.g

.,

–ru

les

abou

t ter

ms

& c

ondi

tions

ass

ocia

ted

with

apr

oduc

t or

serv

ice

in In

tern

et p

urch

asin

g.

–ex

chan

ge a

mon

g m

ultip

le s

uppl

y ch

ain

play

ers:

•pr

ice

vs. q

uant

ity v

s. d

eliv

ery

date

•w

hen

and

how

to o

rder

or

retu

rn it

ems,

that

impa

ctpl

anni

ng.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ru

les:

F

un

da

me

nta

l Te

chn

ica

l Ap

pro

ach

(co

ntin

ue

d)

•D

ecla

rativ

e ap

proa

ch:

prov

ide se

man

tics th

at is

cle

an a

nd d

eep.

•F

acili

tate

spe

cific

atio

n of a

giv

en r

ule

set:

–by

mul

tiple

aut

hors

, cro

ss-e

nter

pris

e, c

ross

-app

licat

ion

–by

non

-tec

hnic

al a

utho

rs

–dy

nam

ical

ly

–w

ith a

bstr

actio

n le

vel m

ore

easi

ly h

uman

-und

erst

anda

ble

•E

nabl

e con

flict

han

dlin

g in

mul

tiple

rul

e sy

stem

s.

•In

tero

pera

te bet

wee

n m

ultip

le r

ule

syst

ems

via

com

mon

-cor

ein

terli

ngua

: in

ter-

agen

t sta

ndar

ds in

X

ML

.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ap

plic

atio

ns

of R

ule

s o

ur

wo

rk to

-da

te: O

verv

iew

•C

an v

iew

gen

eral

ly in

term

s of

bus

ines

s pr

oces

ses,

incl

udin

g w

orkf

low

.

–ru

les

are

good

to c

aptu

re if

-the

n c

ondi

tiona

lity e

sp. i

nvol

ving

cha

inin

g.

•S

tore

fron

t/cat

alog

-bas

ed s

ervi

ces:

ini

tially

, B2C

per

sona

lizat

ion/

prom

otio

ns.

•*C

ontr

acts

/agr

eem

ents

.

–re

pres

ent p

rodu

cts

or s

ervi

ces,

or

serv

ice

term

s &

con

ditio

ns o

f pro

duct

.

–ex

ecut

able s

peci

ficat

ion

–pa

rtia

lly-s

peci

fied

/ tem

plat

e, d

urin

g pr

oces

s of

neg

otia

tion.

•*N

egot

iatio

n, es

p. B

2B:

–re

pres

ent c

onte

nts

of p

ropo

sals

, cou

nter

-pro

posa

ls,

RF

Q’s,

RF

P’s.

–co

nfig

ure

auct

ion

mec

hani

sms

base

d on

con

trac

t tem

plat

es.

•*S

ecur

ity a

utho

rizat

ion

polic

ies:

inc

ludi

ng d

eleg

atio

n, c

ertif

icat

es.

–of

ten,

are

rea

lly a

par

t of o

vera

ll bu

sine

ss p

olic

y, a

t app

licat

ion-

leve

l.

•*

= in

EE

CO

MS

inte

r-en

terp

rise

supp

ly c

hain

sce

nario

s. E

EC

OM

S is

a$2

9Mill

ion

3-ye

ar N

IST

AT

P c

onso

rtiu

m e

ffort

led

by IB

M.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ap

plic

atio

ns

of R

ule

s:e

arlie

r w

ork

on

Ag

en

t B

uild

ing

En

viro

nm

en

t

•C

an v

iew

gen

eral

ly in

term

s of

bus

ines

s pr

oces

ses,

incl

udin

g w

orkf

low

.

–ru

les

are

good

to c

aptu

re if

-the

n c

ondi

tiona

lity e

sp. i

nvol

ving

cha

inin

g.

•E

mbe

ddab

le te

chno

logy

for

build

ing

rule

-bas

ed in

telli

gent

age

nt c

apab

ilitie

sin

to a

pplic

atio

ns.

•C

lass

of a

pplic

atio

ns:

filte

ring

and rout

ing

of i

nfo

item

s

–m

ail,

new

s, L

otus

Not

es d

ocum

ents

–cu

stom

er s

ervi

ce /

help

des

k

–w

orkf

low

in m

anuf

actu

ring:

des

ign

chan

ges,

pla

nt fl

oor

aler

ts

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

EE

CO

MS

Exa

mp

le o

f C

on

flict

ing

Ru

les

•V

endo

r’s r

ules

that

pre

scrib

e ho

w b

uyer

mus

t pla

ce o

r m

odify

an

orde

r:

•A

) 14

day

s ah

ead

if th

e bu

yer

is a

qua

lifie

d cu

stom

er.

•B

) 30

day

s ah

ead

if th

e or

dere

d ite

m is

a m

inor

par

t.

•C

) 2

days

ahe

ad if

the

orde

red

item

’s it

em-t

ype

is b

ackl

ogge

d at

the

vend

or,

the

orde

r is

a m

odifi

catio

n to

red

uce

the

quan

tity

of th

e ite

m, a

nd th

e bu

yer

is a

qual

ified

cus

tom

er.

•S

uppo

se m

ore

than

one

of t

he a

bove

app

lies

to th

e cu

rren

t ord

er?

Con

flict

!

•H

elpf

ul A

ppro

ach:

pre

cede

nce b

etw

een

the

rule

s. O

ften

only

p

art

ial o

rder

of

prec

eden

ce is

just

ified

. E

.g.,

C

� A.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

urt

eo

us

LP

’s: E

xam

ple

•<

lead

Tim

eRul

e1> o

rder

Mod

ifica

tionN

otic

e(?O

rder

,14d

ays)

pref

erre

dCus

tom

erO

f(?

Buy

er,?

Sel

ler)

∧•

p

urch

aseO

rder(?

Ord

er,?B

uyer

,?S

elle

r) .

•<

lead

Tim

eRul

e2>

orde

rMod

ifica

tionN

otic

e(?O

rder

,30d

ays)

← m

inor

Par

t(?B

uyer

,?S

elle

r,?O

rder

) ∧•

pur

chas

eOrd

er(?

Ord

er,?B

uyer

,?S

elle

r) .

•<

lead

Tim

eRul

e3>

orde

rMod

ifica

tionN

otic

e(?O

rder

,2da

ys)

pre

ferr

edC

usto

mer

Of

(?B

uyer

,?S

elle

r) ∧

ord

erM

odifi

catio

nTyp

e(?O

rder

,red

uce)

∧•

orde

rIte

mIs

InB

ackl

og(?O

rder

) ∧•

pur

chas

eOrd

er(?

Ord

er,?B

uyer

,?S

elle

r) .

•ov

errid

es(le

adT

imeR

ule3

, le

adT

imeR

ule1

) .

•⊥

← o

rder

Mod

ifica

tionN

otic

e(?O

rder

,?X

) ∧•

ord

erM

odifi

catio

nNot

ice(?

Ord

er,?

Y);

GIV

EN

?X

� ?Y

.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

XM

L In

terlin

gu

a fo

r E

xam

ple

•<c

lp>

• <

erul

e ru

lela

bel="l

eadT

imeR

ule1

">

<he

ad>

<c

liter

al p

redi

cate

="or

derM

odifi

catio

nNot

ice">

<

varia

ble

nam

e="?

Ord

er"/

>

<

func

tion

nam

e="d

ays1

4"/>

<

/clit

eral

>

</h

ead>

<bo

dy>

<

and>

<

fclit

eral

pre

dica

te=

"pref

erre

dCus

tom

erO

f">

<va

riabl

e na

me=

"?B

uyer

"/>

<va

riabl

e na

me=

"?S

elle

r"/>

<

/fclit

eral

>

<

fclit

eral

pre

dica

te=

"purc

hase

Ord

er">

<va

riabl

e na

me=

"?O

rder

"/>

<va

riabl

e na

me=

"?B

uyer

"/>

<va

riabl

e na

me=

"?S

elle

r"/>

<

/fclit

eral

>

<

/and

>

</b

ody>

• <

/eru

le>

•...

•<

/clp

>

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

mm

on

Ru

les E

xam

ple

:b

oo

ksto

re W

eb

sto

refr

on

t

•B

2C p

erso

naliz

ed p

rom

otio

ns:

–di

scou

ntin

g

–sh

owin

g ta

rget

ed a

ds w

ith in

cent

ives

•R

ules

& fa

cts

from

:

–m

arke

ting

man

ager

s: w

ith u

pdat

es &

mer

ges

•pr

iorit

ies

from

rece

ncy,

auth

ority

, spe

cific

ity

–da

ta m

inin

g

–D

B

–dy

nam

ic W

eb s

essi

on d

ata

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

•R

unni

ng e

xam

ple

in Com

mon

Rul

es: i

nclu

des

abou

t 60

rule

s an

d fa

cts.

•S

ee IB

M R

esea

rch

Rep

ort R

C 2

1473

“D

IPLO

MA

T...

Dem

onst

ratio

n”

•A

ltern

ativ

ely,

see

file

boo

ksto

reE

xam

pleU

nifie

d10-

30-9

8.tx

t .

INS

ER

T B

oo

ksto

re W

eb

E-S

tore

fro

nt

Ap

p E

xam

ple

SL

IDE

S

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

mm

on

Ru

les te

chn

olo

gy

ove

rvie

w

•Ja

va li

brar

y, V

1 pr

otot

ype

runn

ing.

–pl

an e

xter

nal a

lpha

rel

ease

7/3

0/99

on

Alp

haW

orks

http

://al

phaw

orks.

ibm

.com

–pi

lotin

g in

EE

CO

MS

$29

Mill

ion

NIS

T A

TP

pro

ject

(IB

M,

Baa

n, B

oein

g, u

nive

rsiti

es, o

ther

co.

’s)

•B

asic

rul

e re

pres

enta

tion:

L

ogic

pro

gram

s (LP

’s).

–LP

’s in

dec

lara

tive

sens

e, n

ot P

rolo

g. E

.g.,

forw

ard

orba

ckw

ard

chai

ning

.

–re

pres

enta

tion

= s

ynta

x +

dee

p se

man

tics.

•se

man

tics

of r

ule

set =

its

set o

f val

id c

oncl

usio

ns.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

mm

on

Ru

les te

chn

olo

gy

ove

rvie

w (

con

tinu

ed

)•

Ext

ends

rul

e re

pres

enta

tion

to:

–C

ourt

eous

LP

’s:

•pr

iorit

ized

han

dlin

g of

conf

licts

, e.g

., in

upd

atin

g/m

ergi

ng.

–S

ituat

ed (

Cou

rteo

us)

LP’s

:

•pr

oced

ural

atta

chm

ents

to in

voke

non

-rea

soni

ng a

ctio

ns o

rqu

erie

s, v

ia m

etho

ds e

xter

nal t

o in

fere

ncin

g eng

ine.

•C

ourt

eous

Com

pile

r from

cou

rteo

us L

P’s

to o

rdin

ary

LP’s

.

•X

ML

Inte

rling

ua a

nd s

ampl

e tran

slat

ors.

–in

terli

ngua

= c

omm

on r

ule

repr

esen

tatio

n fo

r tr

ansl

atio

n be

twee

nhe

tero

gene

ous

rule

sys

tem

s. S

uita

ble

to b

ecom

e in

dust

ry s

tand

ard.

•S

ampl

e Inf

eren

cing

/Exe

cutio

n Eng

ine:

–fo

rwar

d-ch

aini

ng s

ituat

ed c

ourt

eous

LP

’s.

Cu

rre

nt-

vers

ion

Co

mm

on

Ru

les

app

N

app

1

app

2

tran

sfor

mer

inte

rling

ua

pars

ing/t

rans

latin

gin

& o

ut

deep

sha

red

sem

antic

sin

com

mon

repr

esen

tatio

n:

com

mon

cor

esLogi

cP

rogr

am

fam

ily

XR

ule

fam

ily

YR

ule

fam

ily

rule

sys

1

rule

sys

2

rule

sys

N

Het

erog

eneo

us

cour

teou

s

ordi

nary

/van

illa

repr

esen

tatio

n

mut

expr

iorit

ies

repr

esen

tatio

n

equi

vale

nt

sem

antic

ally

KR

obj’s

strin

g

XS

Bfo

rmat

s

Sm

odel

s

CR

.co

urte

ous

conf

lict-

hand

ling

Log.

Pro

g.

situ

ated

cou

rteo

us L

P’s

XM

L,

KIF

,

engi

ne:

forw

ard

situ

ated

LP

rule

sys

tem

s

obje

cts

othe

r

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

•E

.g.,

in O

O a

pp’s

, DB

’s, w

orkf

low

s.

•Lo

gic

Pro

gram

s (in

pur

e kn

owle

dge-

repr

esen

tatio

n se

nse)

: e.

g.,

–R

elat

iona

l dat

abas

es, S

QL.

–P

rolo

g; K

now

ledg

e-ba

sed

syst

ems.

•P

rodu

ctio

n ru

les (O

PS

5 he

ritag

e):

e.g.

,

–N

euro

n D

ata

rule

-bas

ed J

ava

obje

cts.

–IB

M V

isua

lBan

ker u

sing

(H

aley

).

•E

vent

-Con

ditio

n-A

ctio

n ru

les (lo

ose

fam

ily),

cf.:

–bu

sine

ss p

roce

ss a

utom

atio

n / w

orkf

low

tool

s.

–ac

tive

data

base

s; p

ublis

h-su

bscr

ibe.

Fla

vors

of

Ru

les

Co

mm

erc

ially

Mo

stIm

po

rta

nt

in E

-Bu

sin

ess

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ove

rvie

w (

rep

ea

ted

)•

Intr

o &

Bac

kgro

und

–O

vera

ll m

issi

on, t

echn

ical

app

roac

h, in

dust

ry tr

ends

–E

EC

OM

S p

roje

ct (

NIS

T A

TP

) on

inte

r-en

terp

rise

supp

ly c

hain

inte

grat

ion

•C

ore

Tec

hnol

ogy:

Com

mon

Rul

es pr

otot

ype

base

d on

logi

c pr

ogra

m K

R

–In

nova

tive

conf

lict h

andl

ing,

pro

cedu

ral a

ttach

men

ts; s

ampl

e en

gine

–X

ML

Inte

rling

ua b

etw

een

hete

roge

neou

s ru

le s

yste

ms;

sta

ndar

ds p

lay

•A

pplic

atio

ns:

repr

esen

t bus

ines

s pr

oces

ses/

wor

kflo

w, p

olic

ies,

pro

duct

s

–C

ontr

acts

/agr

eem

ents

, neg

otia

tion

esp.

B2B

, sec

urity

aut

horiz

atio

n

–E

EC

OM

S s

uppl

y ch

ain

scen

ario

s, N

et.C

omm

erce

-typ

e B

2C s

cena

rio

•D

irect

ions

–E

xter

nal a

lpha

rel

ease

in 7

/99

of c

ore

tech

nolo

gy

–E

xplo

re m

ore

appl

icat

ions

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

•P

art 1

: fo

r In

trod

ucto

ry s

ectio

n of

talk

•P

art 2

: ab

out E

EC

OM

S

•P

art 3

: m

ore

Tec

hnic

al D

etai

ls, i

nclu

ding

logi

c pr

ogra

ms,

cou

rteo

us, s

ituat

ed,

appl

icat

ions

, etc

..

OU

TL

INE

OF

OP

TIO

NA

LS

SL

IDE

S

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

PA

RT

-1 O

PT

ION

AL

S F

OL

LO

W:

fro

m I

NT

RO

DU

CT

OR

Y-s

ect

ion

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ru

les

acr

oss

Ap

plic

atio

ns

App

licat

ion

1, e

.g.,

sel

ler

e-st

oref

ront

App

licat

ion

2, e

.g.,

buye

r sho

pbot

agen

t

Bus

ines

sLo

gic

Bus

ines

sLo

gic

Rul

esR

ules

Inte

rcha

nge

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ap

plic

atio

n U

sin

g R

ule

s

repr

esen

tatio

n

rule

edito

r

Rul

e S

yste

m

App

licat

ion

rule

exe

cutio

n

engi

ne

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ro

les

for

Ru

les

•1s

t ste

p: R

ules

as

rigor

ous

spec

ifica

tion

with

out e

xecu

tion.

•2n

d st

ep (o

ur m

ain

focu

s): R

ules

as

exec

utab

le s

peci

ficat

ion.

–C

oars

e-gr

ain

inte

grat

ion

with

res

t of c

ode.

E.g

.:

•ru

les

engi

ne in

wra

pper

, age

nt, w

orkf

low

, dat

abas

e, e

xten

ded

tran

sact

ion

mon

itor,

….

•3r

d st

ep:

Fin

e-gr

ain

inte

grat

ion

with

res

t of c

ode.

E.g

.:

–O

bjec

t-or

ient

ed a

pplic

atio

n de

velo

pmen

t too

l’s r

ules

feat

ure.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ru

les

Au

tho

rin

g A

pp

roa

che

s

•G

raph

ical

.

•N

atur

al L

angu

age

(lim

ited)

.

•P

re-d

efin

ed te

mpl

ates

and

voc

abul

ary.

–do

mai

n/in

dust

ry s

peci

fic.

–ap

plic

atio

n/su

ite s

peci

fic.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

PA

RT

-2 O

PT

ION

AL

S F

OL

LO

W:

ab

ou

t E

EC

OM

S

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

EE

CO

MS

Su

pp

ly C

ha

in P

roje

ct: O

verv

iew

•E

EC

OM

S =

Ext

ende

d E

nter

pris

e C

onso

rtiu

m fo

r In

tegr

ated

Col

labo

rativ

eM

anuf

actu

ring

Sys

tem

s.

•IB

M-le

d co

nsor

tium

, 50%

-fun

ded

by U

S g

over

nmen

t’s N

IST

Adv

ance

dT

echn

olog

y P

rogr

am.

$29M

ove

r 3

year

s, e

nds

2001

.

•A

dvan

ced

supp

ly c

hain

man

agem

ent.

Con

sort

ium

incl

udes

man

ufac

turin

gso

ftwar

e ve

ndor

s, r

ules

and

tool

ven

dors

, man

ufac

ture

r cu

stom

ers.

•B

usin

ess

Foc

us:

impr

ove “a

gilit

y” o

f man

ufac

turin

g. R

espo

nd to

com

mon

but u

npre

dict

able

eve

nts

such

as

late

del

iver

y, p

lant

line

bre

akdo

wn,

larg

erth

an e

xpec

ted

orde

r. R

eact

qui

ckly

, inc

ludi

ng m

odify

pla

ns, s

ched

ules

.In

tegr

ate:

typ

ical

ly m

ulti-

appl

icat

ion,

ver

y of

ten

mul

ti-en

terp

rise.

•T

echn

ical

Foc

us: ru

les

and

conf

lict h

andl

ing;

virt

ual s

ituat

ion

room

for

hum

an c

olla

bora

tive

wor

kflo

w; a

ttend

ant t

ools

, age

nts,

and

sec

urity

issu

es.

•Is

follo

w-o

n to

CIIM

PLE

X (

IBM

-led

NIS

T A

TP

$22

M)

& c

halle

nges

itid

entif

ied.

S

hare

s: c

onso

rtiu

m, s

cena

rios,

age

nt-b

ased

app

roac

h.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

EE

CO

MS

Su

pp

ly C

ha

in P

roje

ct: Our

Rol

eO

ur R

ole

•R

ules

prim

arily

:

–ap

plic

atio

n S

cena

rios

with

cus

tom

ers.

–in

terli

ngua

.

–co

nflic

t han

dlin

g.

–fo

r se

curit

y au

thor

izat

ion

polic

ies

(tru

st m

anag

emen

t).

•M

ultip

le, h

eter

ogen

eous

rul

e sy

stem

s w

ithin

con

sort

ium

: fr

omve

ndor

s, u

nive

rsiti

es.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

EE

CO

MS

: P

art

icip

an

ts

•IB

M is

lead

er o

f con

sort

ium

, via

its

Man

ufac

turin

g IS

U.

Run

s E

EC

OM

SP

roje

ct O

ffice

, EE

CO

MS

Inte

grat

ion

Cen

ter

and

Too

ls D

ev..

•IB

M o

rg.’s

invo

lved

: T

.J. W

atso

n R

esea

rch,

Sup

ply

Cha

in O

ptim

izat

ion

Pro

d.D

ev.,

Sup

ply

Cha

in Man

uf. S

ol’n

s, S

ynch

rono

us C

olla

bora

tion

Too

ls a

nd A

D(A

ustin

).

•B

oein

g m

ilita

ry a

ircra

ft. A

maj

or m

anuf

actu

rer.

Act

s as

cus

tom

er.

•B

ercl

ain,

par

t of B

aan:

maj

or m

anuf

actu

ring

/ ER

P s

oftw

are

vend

or.

•U

nive

rsiti

es a

s IB

M s

ub-c

ontr

acto

rs:

U. N

C C

harlo

tte, U

. MD

Bal

timor

e, U

.F

lorid

a. C

ontr

ibut

e re

sear

ch-

ily t

o al

l tec

hnic

al fo

ci.

•V

itria

, Env

isio

nIt,

IndX

, Sca

ndur

a: (s

mal

l) to

ols/

spec

ifica

tion

vend

ors.

•T

RW

sup

ply-

chai

n co

nsul

ting

prac

tice.

Act

s as

cus

tom

er(s

).

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

EE

CO

MS

: M

ain

Co

llab

ora

tors

•P

rof.’

s T

im F

inin

and

Yan

nis

Labr

ou o

f U. M

aryl

and

(Bal

timor

e):

–ne

gotia

tion

prot

ocol

s be

twee

n ag

ents

; FIP

A s

tand

ards

for

thes

e

–X

ML

vers

ions

of F

IPA

and

KIF

dra

ft st

anda

rds

–se

curit

y au

thor

izat

ion

polic

ies,

del

egat

ion

•P

rof.’

s B

ill C

hu a

nd B

ob W

ilhel

m o

f U. N

. Car

olin

a (C

harlo

tte)

–co

nstr

aint

sat

isfa

ctio

n an

d op

timiz

atio

n in

priv

ate

nego

tiatio

n de

cisi

ons

–in

tegr

atin

g ru

le-b

ased

aut

horiz

atio

n po

licie

s in

to s

ecur

ity s

ervi

ces

•P

rof.’

s S

tanl

ey S

u an

d Jo

achi

m H

amm

er o

f U. F

lorid

a (G

aine

svill

e)

–ne

gotia

tion

arch

itect

ure

–E

vent

-Con

ditio

n-A

ctio

n ru

le s

yste

m

•Ly

nne

Thi

eme o

f Vitr

ia T

echn

olog

ies

–bu

sine

ss p

roce

ss a

utom

atio

n an

d pu

blis

h-su

bscr

ibe

–E

vent

-Con

ditio

n-A

ctio

n ru

le s

yste

m

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

EE

CO

MS

: V

irtu

al S

itua

tion

Ro

om

•H

uman

bus

ines

s pr

oces

s an

d w

orkf

low

: fo

r tim

ely,

agi

le r

espo

nse.

•D

istr

ibut

ed, i

nter

-ent

erpr

ise

.

•Le

ads:

–C

once

pt:

Bill

Tol

one

of U

NC

Cha

rlotte

(IB

M s

ub-c

ontr

acto

r)

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

PA

RT

-3 O

PT

ION

AL

S F

OL

LO

W:

ab

ou

t m

ore

Te

chn

ica

l De

tails

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Lo

gic

Pro

gra

ms

as

ba

sic

rep

rese

nta

tion

: D

efin

itio

n•

A L

P is

a s

et o

f (pr

emis

e) r

ules

; sem

antic

ally

, it s

peci

fies

a se

t of c

oncl

usio

ns.

•E

xam

ple

rule

:

• s

endP

age(?

msg

,Joe

) ← fr

om(?

msg

,?s)

∧ ur

gent

(?msg

) ∧

care

sAbo

ut(Jo

e,?s

).

•w

here

the

“?”

pref

ix in

dica

tes

a lo

gica

l var

iabl

e.

•G

ener

ally

, a r

ule

has

the

form

of

Hea

d IF

Bod

y

:

H

B_1

∧ ..

. ∧ B

_j ∧

~B

_j+

1 ∧

... ∧

~B

_m .

•w

here

m ≥

0 ;

∧ st

ands

for

logi

cal “A

ND

”; ←

sta

nds

for

logi

cal “IF

”; a

ndH

, B_1

, ...,

B_m

ar

e ea

ch a

n atom

with

form

: P

redi

cate

(Ter

m_1 , ..

., T

erm

_k ).

•A

pre

dica

te =

a r

elat

ion.

An

atom

sem

antic

ally

den

otes

a b

oole

an.

•~

sta

nds

for n

egat

ion-

as-f

ailu

re (a.

k.a.

wea

k ne

gatio

n, d

efau

lt ne

gatio

n).

–T

he n

egat

ion-

as-f

ailu

re c

onst

ruct

is lo

gica

lly n

on-m

onot

onic

.

–In

tuiti

vely

, ~p

mea

ns p’s

trut

h va

lue

is e

ither

fals

e O

R u

nkn

ow

n .

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Lo

gic

Pro

gra

ms:

D

efin

itio

n(c

on

tinu

ed

)

•E

ach

argu

men

t Ter

m_

1 , ..

., T

erm

_k is

a te

rm.

•A

term

is e

ither

a logi

cal c

onst

ant (e.

g., “

Joe”

) O

R a

logi

cal v

aria

ble (

e.g.

,“?

msg

”) O

R a

func

tiona

l exp

ress

ion o

f the

form

:

• L

ogic

alF

unct

ion(T

erm

_1, .

.., T

erm

_k )

•A

func

tiona

l exp

ress

ion

sem

antic

ally

ess

entia

lly d

enot

es a

logi

cal c

onst

ant.

•A

term

, ato

m, o

r ru

le is

cal

led “g

roun

d” w

hen

it ha

s no

logi

cal v

aria

bles

.

•A

fac

t is

a gr

ound

rul

e w

ith e

mpt

y bo

dy.

•A

prim

itive

con

clus

ion h

as th

e fo

rm o

f a g

roun

d at

om (

com

poun

d co

nclu

sion

sar

e bu

ilt u

p fr

om th

ese

via

logi

cal o

pera

tors

suc

h as

AN

D e

tc.)

.

•S

eman

tical

ly, a

rul

e or

LP

sta

nds

for

the

set o

f all

its g

roun

d in

stan

ces.

•(O

bser

ve th

at a

rul

e bo

dy c

an r

epre

sent

an

expr

essi

on in

rel

atio

nal a

lgeb

ra

cf.

rela

tiona

l DB

’s (

e.g.

, SQ

L).)

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Lo

gic

Pro

gra

ms

as

ba

sic

rep

rese

nta

tion

: A

dva

nta

ge

s

•D

ecla

rativ

e: s

eman

tics

is in

depe

nden

t of

infe

renc

ing p

roce

dure

impl

emen

tatio

n, e

.g.,

forw

ard

vs

. bac

kwar

d ch

aini

ng, s

eque

ncin

g of

exec

utin

g ru

les

or c

ondi

tions

with

in r

ules

.

•E

xpre

ssiv

e: r

elat

iona

l exp

ress

ions

cf

. SQ

L, la

rge

frag

men

t of f

irst-

orde

r lo

gic,

chai

ning

, bas

ic lo

gica

l non

-mon

oton

icity

(un

like

first

-or

der

logi

c).

•E

ffici

ent:

com

puta

tiona

lly tr

acta

ble

give

n tw

o re

ason

able

res

tric

tions

:–

1. D

atal

og =

no

logi

cal f

unct

ions

of n

on-z

ero ar

ity.

–2.

Bou

nded

num

ber

v o

f log

ical

var

iabl

es p

er r

ule.

–m

= O

( n̂

(v+

1) ),

whe

re n

= ||

LP||,

m =

||gr

ound

-inst

antia

ted

LP||.

–In

fere

ncin

g tim

e is

O

( m

) fo

r br

oad

(acyc

lic)

case

, O

(m2̂ )

gen

eral

ly(f

or w

ell-f

ound

ed s

eman

tics)

.

–B

y co

ntra

st, f

irst-

orde

r-lo

gic in

fere

ncin

g is

NP

-har

d.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Lo

gic

Pro

gra

ms:

A

dva

nta

ge

s(c

on

tinu

ed

)

•W

idel

y de

ploy

ed a

nd fa

mili

ar:

–re

latio

nal D

B’s

, SQ

L

–P

rolo

g

–in

telli

gent

age

nts

and

know

ledg

e-ba

sed

syst

ems

•e.

g., I

BM

’s A

gent

Bui

ldin

g E

nviro

nmen

t

•C

omm

on c

ore

shar

ed s

eman

tical

ly b

y m

any

rule

sys

tem

s:

e.g.

,pr

oduc

tion

rule

s, E

vent

-Con

ditio

n-A

ctio

n ru

les,

firs

t-or

der-

logi

c ea

chov

erla

p st

rong

ly w

ith L

P’s

.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

urt

eo

us

LP

’s: th

e W

ha

t

•U

pdat

ing/

mer

ging

of r

ules

set

s: i

s cr

ucia

l, of

ten

gene

rate

s co

nflic

t.

•C

ourt

eous

LP’s

feat

ure

prio

ritiz

ed h

andl

ing

of c

onfli

cts.

•S

peci

fy s

cope

of c

onfli

ct v

ia a

set

of

mut

ual e

xclu

sion

cons

trai

nts.

–E

.g.,

⊥ ←

dis

coun

t(?p

rodu

ct,5

%) ∧

disc

ount

(?pr

oduc

t,10%

) .

–E

.g.,

⊥ ←

loya

lCus

tom

er(?c

,?s)

∧ pr

emie

reC

usto

mer(?

c,?s

) .

–P

erm

it cla

ssic

al-n

egat

ion o

f ato

ms:

¬p

mea

ns p

has

trut

h va

lue

fals

e

•im

plic

itly,

← p

∧ ¬

p

for

ever

y at

om p

.

•P

riorit

ies

bet

wee

n ru

les:

part

ially

-ord

ered

.

–R

epre

sent

prio

ritie

s vi

a rese

rved

pre

dica

te that

com

pare

s rule

labe

ls:

•ov

errid

es(r

ule1

,rul

e2)

m

eans

rul

e1 is

hig

her-

prio

rity

than

rul

e2.

•E

ach

rule

opt

iona

lly h

as a

rul

e la

bel w

hose

form

is a

logi

cal c

onst

ant.

•ov

errid

es c

an b

e re

ason

ed a

bout

, jus

t lik

e an

y ot

her

pred

icat

e.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

urt

eo

us

LP

’s: E

xam

ple

(re

pe

ate

d)

•<

lead

Tim

eRul

e1> o

rder

Mod

ifica

tionN

otic

e(?O

rder

,14d

ays)

pref

erre

dCus

tom

erO

f(?

Buy

er,?

Sel

ler)

∧•

p

urch

aseO

rder(?

Ord

er,?B

uyer

,?S

elle

r) .

•<

lead

Tim

eRul

e2>

orde

rMod

ifica

tionN

otic

e(?O

rder

,30d

ays)

← m

inor

Par

t(?B

uyer

,?S

elle

r,?O

rder

) ∧•

pur

chas

eOrd

er(?

Ord

er,?B

uyer

,?S

elle

r) .

•<

lead

Tim

eRul

e3>

orde

rMod

ifica

tionN

otic

e(?O

rder

,2da

ys)

pre

ferr

edC

usto

mer

Of

(?B

uyer

,?S

elle

r) ∧

ord

erM

odifi

catio

nTyp

e(?O

rder

,red

uce)

∧•

orde

rIte

mIs

InB

ackl

og(?O

rder

) ∧•

pur

chas

eOrd

er(?

Ord

er,?B

uyer

,?S

elle

r) .

•ov

errid

es(le

adT

imeR

ule3

, le

adT

imeR

ule1

) .

•⊥

← o

rder

Mod

ifica

tionN

otic

e(?O

rder

,?X

) ∧•

ord

erM

odifi

catio

nNot

ice(?

Ord

er,?

Y);

GIV

EN

?X

� ?Y

.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Prio

ritie

s a

re a

vaila

ble

an

d u

sefu

l

•P

riorit

y in

form

atio

n is

nat

ural

ly a

vaila

ble

and

usef

ul.

E.g

.,–

rece

ncy:

hig

her

prio

rity

for

mor

e re

cent

upd

ates

.

–sp

ecifi

city

: hi

gher

prio

rity

for

mor

e sp

ecifi

c ca

ses

(e.g

., ex

cept

iona

l cas

es,

sub-

case

s, in

herit

ance

).

–au

thor

ity:

high

er p

riorit

y fo

r m

ore

auth

orita

tive

sour

ces

(e.g

., le

gal

regu

latio

ns, o

rgan

izat

iona

l im

pera

tives

).

–re

liabi

lity:

hig

her

prio

rity

for

mor

e re

liabl

e so

urce

s (e

.g.,

secu

rity

cert

ifica

tes

obta

ined

by

dele

gatio

n, a

ssum

ptio

ns, o

bser

vatio

nal d

ata)

.

–cl

osed

wor

ld:

low

est p

riorit

y fo

r ca

tch-

case

s.

•M

any

prac

tical

rul

e sy

stem

s em

ploy

prio

ritie

s of

som

e ki

nd, o

ften

impl

icit,

e.g

.,–

rule

seq

uenc

ing

in P

rolo

g an

d pr

oduc

tion

rule

s.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

urt

eo

us

LP

’s: A

dva

nta

ge

s

•F

acili

tate

upd

atin

g an

d m

ergi

ng.

•E

xpre

ssiv

e: c

lass

ical

neg

atio

n, m

utua

l exc

lusi

ons,

par

tially

-ord

ered

prio

ritiz

atio

n, r

easo

ning

to in

fer

prio

ritiz

atio

n.

•G

uara

ntee

cons

iste

nt, u

niqu

e set

of c

oncl

usio

ns.

–M

utua

l exc

lusi

on is

enf

orce

d . E

.g.,

neve

r co

nclu

de b

oth

p &

¬p.

•E

ffici

ent:

low

com

puta

tiona

l ove

rhea

d be

yond

ord

inar

y LP’s.

–T

ract

able

giv

en r

easo

nabl

e re

stric

tions

(D

atal

og, b

ound

v o

n #va

r’s/r

ule)

:

•ex

tra

cost

is e

quiv

alen

t to

incr

easi

ng v

to (

v+2)

in o

rdin

ary

LP’s

.

–B

y co

ntra

st, m

ore

expr

essi

ve p

riorit

ized

rul

e re

pres

enta

tions

(e.

g.,

Prio

ritiz

ed D

efau

lt Lo

gic)

add

NP

-har

d ov

erhe

ad.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Situ

ate

d L

P’s

: O

verv

iew

•P

oint

of d

epar

ture

: LP

’s a

re

pure

-bel

ief rep

rese

ntat

ion,

but

mos

tpr

actic

al r

ule

syst

ems

wan

t to

invo

ke e

xter

nal p

roce

dure

s.

•S

ituat

ed L

P ‘s

feat

ure

a se

man

tical

ly-

clea

n ki

nd o

f pro

cedu

ral

atta

chm

ents .

I.e

., th

ey ho

ok b

elie

fs to

driv

e pr

oced

ural

AP

I’s

outs

ide

the

rule

eng

ine.

•P

roce

dura

l atta

chm

ents

for

sens

ing (

quer

ies)

whe

n te

stin

g an

ante

cede

nt c

ondi

tion

or fo

r effe

ctin

g (a

ctio

ns)

upon

con

clud

ing

aco

nseq

uent

con

ditio

n. A

ttach

ed p

roce

dure

is in

voke

d w

hen

test

ing

orco

nclu

ding

in in

fere

ncin

g.

•S

enso

r or

effe

ctor

link

sta

tem

ent s

peci

fies

an a

ssoc

iatio

n fr

om a

pred

icat

e to

a p

roce

dura

l cal

l pat

tern

, e.g

., a

met

hod.

A

link

issp

ecifi

ed a

s par

t of t

he r

epre

sent

atio

n.

I.e.,

a S

LP is

a cond

uct s

et tha

tin

clud

es li

nks

as w

ell a

s ru

les.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Situ

ate

d L

P’s

: O

verv

iew

(co

ntin

ue

d)

• p

hone

Num

berO

fPre

dica

te ::s

:: B

oein

gBlu

ePag

esC

lass.g

etP

hone

Met

hod .

ex.

se

nso

r lin

k

• s

houl

dSen

dPag

ePre

dica

te

::e::

AT

TP

ager

Cla

ss.go

Pag

eMet

hod

.e

x. e

ffe

cto

r lin

k

•S

enso

r pr

oced

ure

may

req

uire

som

e ar

gum

ents

to b

e gr

ound

, i.e

.,bo

und;

in g

ener

al it

has

a s

peci

fied

bind

ing-

sign

atur

e.

•E

nabl

e dyn

amic

load

ing a

nd re

mot

e lo

adin

g of t

he a

ttach

ed p

roce

dure

s(e

xplo

it Ja

va g

oodn

ess)

.

•O

vera

ll: c

lean

ly s

epar

ate

out t

he p

roce

dura

l sem

antic

s as

a d

ecla

rativ

eex

tens

ion

of th

e pu

re-b

elie

f dec

lara

tive

sem

antic

s. E

asily

sep

arat

ech

aini

ng fr

om a

ctio

n.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

urt

eo

us

Co

mp

iler

•T

rans

form

er co

mpi

les

a co

urte

ous

LP in

to a

n or

dina

ry L

P.

•A

rad

ical

ly in

nova

tive

appr

oach

in r

ules

rep

rese

ntat

ion.

•“C

ompi

les

away

” co

nflic

t, as

mod

ular

add

-on to

rul

esy

stem

X’s

–in

fere

ncin

g

–sp

ecifi

catio

n

•E

nabl

es c

ourt

eous

feat

ures

to b

e ad

ded

to, o

r im

plem

ente

din

, a v

arie

ty o

f rul

e sy

stem

s.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Inte

rlin

gu

a C

on

cep

t

•C

halle

nge:

het

erog

enei

ty o

f rul

e sy

stem

s (w

ithin

appl

icat

ions

) to

be

inte

grat

ed.

–E

ach

has

own

rule

rep

rese

ntat

ion.

N o

f the

m.

–R

epre

sent

atio

n =

syn

tax

+

sem

antic

s .•

App

roac

h: t

rans

late

via in

terli

ngua

rep

’n.

–re

p’n

A ↔

inte

rling

ua ↔

rep

’n B

.

–A

dvan

tage

: O

(N)

tran

slat

ors

inst

ead

of O

(N^

2 ).

•*F

ocus

: co

mm

erci

ally

impo

rtan

t rep’

n fa

mili

es:

–LP

’s, p

rodu

ctio

n (O

PS

5), E

vent

-Con

ditio

n-A

ctio

n.

–E

.g.,

in O

O a

pp’s

, DB

’s, w

orkf

low

s.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Inte

rlin

gu

a :D

ee

p S

ha

red

Se

ma

ntic

s a

t C

ore

•D

esire

: de

ep s

eman

tics

(mod

el-t

heor

etic

) to

–un

ders

tand

and

exe

cute im

port

ed r

ules

.

•P

ossi

ble

only

for

shar

ed e

xpre

ssiv

e su

bset

s: “

core

s”.

–R

est t

rans

late

d w

ith s

uper

ficia

l sem

antic

s.

•D

esid

erat

um: d

ecla

rativ

enes

s of c

ore

/ rep

’n (

in s

ense

of k

now

ledg

ere

pres

enta

tion

theo

ry).

–M

axim

ize

over

all a

dvan

tage

s of

rul

es:

•N

on-p

rogr

amm

ers

unde

rsta

nd &

mod

ify.

•D

ynam

ical

ly (

run-

time)

mod

ify.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Inte

rlin

gu

a :G

oin

g B

eyo

nd

KIF

•P

oint

of d

epar

ture

is K

IF:

Kno

wle

dge

Inte

rcha

nge

For

mat

–In

tent

: ge

nera

l-kno

wle

dge in

terli

ngua

.

–E

mer

ging

sta

ndar

d, in

AN

SI co

mm

mitt

ee.

–M

ain

focu

s: c

lass

ical

logi

c,

esp.

firs

t-or

der.

–H

as m

ajor

lim

itatio

ns r

emed

ied

by o

ur

Inte

rling

ua:

•lo

gica

lly m

onot

onic

; no c

onfli

ct h

andl

ing

or p

riorit

ies.

•pu

re-b

elie

f: no

pro

cedu

ral a

ttach

men

ts.

•O

ur In

terli

ngua

use

s si

tuat

ed c

ourt

eous

LP

’s

rep’

n.–

Com

plem

ents

KIF.

O

verla

ps o

n:

•pu

re-b

elie

f rul

es w

ithou

t neg

atio

n-as

-fai

lure

.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

•X

ML

form

at ↔

Jav

a ob

ject

s ↔ t

ext s

ynta

x fo

r ru

le s

yste

m X

•S

ampl

e tra

nsla

tors

to/fr

o K

IF a

nd 3

rul

e sy

stem

s in L

P fa

mily

,in

itial

ly w

ith p

ure-

belie

f sem

antic

s as

cou

rteo

us L

P

rep’

n.

–ba

ckw

ard-d

irect

ion:

XS

B (S

tony

broo

k, co

mm

erci

aliz

ing)

.

–fo

rwar

d-di

rect

ion:

Sm

odel

s (H

elsi

nki,

acad

emic

).

– C

omm

onR

ules

cour

teou

s LP

text

syn

tax:

•co

urte

ous

expr

essi

vene

ss:

mut

ex’s,

prio

ritie

s.

Inte

rlin

gu

a: cu

rre

nt ve

rsio

n

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

•F

IPA

= F

ound

atio

n fo

r In

telli

gent

Phy

sica

l Age

nts

–S

tand

ards

bod

y: m

ain

indu

stry

locu

s of

act

ion

sinc

e ‘9

6 fo

rin

telli

gent

age

nts

know

ledg

e-in

terc

hang

e st

anda

rds

wor

k, e

.g.,

for

Inte

rnet

e-c

omm

erce

ne

gotia

tion b

etw

een

agen

ts.

–C

urre

nt d

raft

stan

dard

use

s A

NS

I KIF

to r

epre

sent

rul

es.

–In

col

labo

ratio

n w

ith U

. Mar

ylan

d, we

are

driv

ing

XM

L ve

rsio

n of

KIF

, in

sync

h w

ith o

ur IB

M r

ules

inte

rling

ua a

nd E

EC

OM

S.

–In

ear

ly d

iscu

ssio

ns a

bout

goi

ng b

eyon

d K

IF.

•A

NS

I KIF

–W

e ha

ve lo

ng-r

unni

ng r

ole

in c

omm

ittee

: n

on-m

onot

onic

ity a

ndco

nflic

t han

dlin

g.

Inte

rlin

gu

a:e

arly

sta

nd

ard

s e

ng

ag

em

en

ts

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Sa

mp

le R

ule

En

gin

e:

curr

en

t ve

rsio

n

•S

ampl

e Inf

eren

cing

/Exe

cutio

n Eng

ine:

–fo

rwar

d-ch

aini

ng s

ituat

ed c

ourt

eous

LP

’s.

•C

ompo

ses c

ou

rte

ou

s co

mp

iler w

ith e

ngin

e fo

r si

tuat

ed

ord

ina

ry L

P’s

.

•In

tent

is p

roof

-of-

conc

ept f

or Inte

rling

ua, c

ourt

eous

com

pile

r, s

ituat

edte

chni

ques

.–

Not

per

form

ance

-tun

ed.

–La

cks

vario

us fe

atur

es fo

und

in b

est-

of-b

reed

com

mer

cial

rul

e sy

stem

s.

•R

estr

icte

d to

acyc

lic c

ase

(no

pred

icat

e de

pend

s th

ru r

ules

on

itse

lf).

•IB

M is

not

in th

e ru

le-e

ngin

e bu

sine

ss, w

e w

ill p

artn

er w

ithco

mm

erci

al r

ule-

syst

em v

endo

rs to

lice

nse

thei

r te

chno

logy

–...

whi

ch C

omm

onR

ules

com

plem

ents

and

ext

ends

.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Co

ntr

act

s/A

gre

em

en

ts

•U

se In

terli

ngua

to r

epre

sent

pro

duct

s or

ser

vice

s, o

r se

rvic

e te

rms

&co

nditi

ons

of p

rodu

ct, e

.g.,

in c

atal

og o

r du

ring

nego

tatio

n.

–E

.g.,

busi

ness

pro

cess

to r

etur

n an

item

for

repa

ir, o

r to

del

iver

an

orde

r.

•E

xecu

tabl

e sp

ecifi

catio

n; situ

ated

/ pr

oced

ural

atta

chm

ents

is

esp.

usef

ul.

•P

artia

lly-s

peci

fied

/ tem

plat

e, es

p. du

ring

proc

ess

of n

egot

iatio

n.

•C

ompl

emen

t XM

L ont

olog

ies a

lread

y ev

olvi

ng fo

r va

rious

dom

ains

.

–O

ntol

ogy

= fo

rmal

ly-r

epre

sent

ed v

ocab

ular

y / d

efin

ition

s

•C

onfig

ure

auto

mat

ed a

uctio

n m

echa

nism

s, b

ased

on

cont

ract

tem

plat

es.

–w

hich

goo

ds, w

hich

attr

ibut

es (

e.g.

, pric

e, d

eliv

ery-

date

) ar

e at

issu

e.

•In

col

labo

ratio

n w

ith:

EE

CO

MS

and

–P

rof.

Mik

e W

ellm

an a

nd P

hD s

tude

nt D

an R

eeve

s of

U. M

ichi

gan

(Ann

Arb

or),

as

IBM

-fun

ded

Uni

vers

ity P

artn

ersh

ip p

roje

ct s

tart

ed 1

/99.

•M

ike

lead

s Auc

tionB

ot, a

maj

or e

xist

ing

auto

mat

ed In

tern

et s

ervi

ce.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Ne

go

tiatio

n, e

sp. B

2B

•U

se In

terli

ngua

to r

epre

sent

con

tent

s of

:

–R

eque

sts

For

Quo

tatio

n or

Pro

posa

l, i.e

., st

atem

ents

of b

uyer

inte

rest

s,th

at in

itiat

e in

ter-

ente

rpris

e ne

gotia

tion.

–re

spon

ses

to s

uch

RF

Q’s

/ RF

P’s

by

selle

r.

–pr

opos

als

and

coun

ter-

prop

osal

s an

d “s

ide

info

rmat

ion”

exc

hang

ed d

urin

gba

ck-a

nd-f

orth

nego

tatio

n / b

arga

inin

g be

twee

n bu

yer

and

selle

r.

–st

atem

ents

of s

elle

r/su

pplie

r ca

pabi

litie

s/in

tere

sts,

e.g

., im

port

ant f

orso

urce

sel

ectio

n as

wel

l as

barg

aini

ng.

•C

onfig

ure

auct

ion

mec

hani

sms

base

d on

con

trac

t tem

plat

es. (

with

U.

Mic

h.)

•In

col

labo

ratio

n w

ith U

. Mic

h. a

nd:

–in

EE

CO

MS

ove

rall

effo

rt,

nego

tiatio

n is

a m

ain

‘99

focu

s of

dem

os a

ndsc

enar

ios.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Se

curity

Au

tho

riza

tion

Po

licie

s

•U

se ru

le-b

ased

exe

cuta

ble

spec

ifica

tion

of s

ecur

ity a

utho

rizat

ion

polic

ies,

a.k.

a. trus

t man

agem

ent

: in

clud

ing

dele

gatio

n, c

ertif

icat

es.

–W

e ha

ve th

e fir

st s

tep

of a

n expr

essi

ve e

xten

sion

of c

ourt

eous

LP

’s to

hand

le d

eleg

atio

n and

cer

tific

ates

.

•O

ften,

aut

horiz

atio

n po

licy

is r

eally

a p

art o

f ove

rall

busi

ness

pol

icy,

at a

pp-

leve

l. T

his

cont

rast

s w

ith a

uthe

ntic

atio

n.

•A

dvan

tage

s of

rul

e-ba

sed

appr

oach

, es

p. fr

om d

ecla

rativ

e se

man

tics:

–pr

inci

pled

han

dlin

g of

neg

atio

n an

d co

nflic

t.

–pr

ovab

le g

uara

ntee

s of

beh

avio

r of

im

plem

enta

tion.

–m

ore

hum

an-u

nder

stan

dabl

e an

d ea

sy to

mod

ify.

–ea

sier

inte

grat

ion

with

gen

eral

bus

ines

s po

licy.

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

Se

curity

Au

tho

riza

tion

Po

licie

s(c

on

tinu

ed

)

•In

col

labo

ratio

n w

ith:

–Jo

an F

eige

nbau

m of A

TT

Res

earc

h w

hich

has

a le

adin

g ef

fort

on

trus

tm

anag

emen

t, an

d he

r P

hD s

tude

nt N

ingh

ui L

i of N

YU

.

–E

EC

OM

S:

as a

mai

n fo

cus

of E

EC

OM

S in

nova

tion

in S

ecur

ity S

IG (

area

of c

onso

rtiu

m).

9/22

/99

by B

enja

min

Gro

sof

IB

M

copy

right

s re

serv

ed

ove

rall

key

Ch

alle

ng

es

ide

ntif

ied

•E

xcha

nge

& u

pdat

e bu

sine

ss r

ules,

dyna

mic

ally

. E

.g.,

–co

mpo

se a

n ap

plic

atio

n fr

om m

ultip

le c

ompo

nent

s.

•N

et.C

omm

erce

-typ

e B

2C:

mar

ketin

g m

anag

er, d

ata

min

er, .

..

–bu

y/se

ll or

inte

grat

e be

twee

n hete

roge

nous p

eer

appl

icat

ions

.

•B

2B in

ter-

ente

rpris

e su

pply

cha

in in

man

ufac

turin

g in

dust

ry.

–E

.g.,

whe

n an

d ho

w to

ord

er o

r re

turn

item

s, th

at im

pact

pla

nnin

g.

•In

tern

et p

urch

asin

g: t

erm

s &

con

ditio

ns.

•F

acili

tate

spe

cific

atio

n, d

ynam

ical

ly, o

f a g

iven

rul

e se

t by

–m

ultip

le a

utho

rs.

•e.

g., c

ross

-ent

erpr

ise

or c

ross

-app

licat

ion.

–no

n-te

chni

cal/n

on-p

rogr

amm

er a

utho

rs.

•m

ake

abst

ract

ion

leve

l mor

e ea

sily

hum

an-u

nder

stan

dabl

e.