+ All Categories
Home > Documents > Ontology Alignment state of the art and an application in literature...

Ontology Alignment state of the art and an application in literature...

Date post: 10-Apr-2018
Category:
Upload: phamkhue
View: 213 times
Download: 0 times
Share this document with a friend
88
Ontology Alignment state of the art and an application in literature search Patrick Lambrix Linköpings universitet
Transcript

Ont

olog

yA

lignm

ent

stat

eof

the

art

and

an a

pplic

atio

nin

lite

ratu

rese

arch

Patr

ick

Lam

brix

Lin

köpi

ngs

univ

ersi

tet

Ont

olog

ies

“Ont

olog

ies

defi

ne th

e ba

sic

term

s an

d re

lati

ons

com

pris

ing

the

voca

bula

ry o

f a

topi

c ar

ea, a

s w

ell a

s th

e ru

les

for

com

bini

ng te

rms

and

rela

tion

s to

def

ine

exte

nsio

ns to

the

voca

bula

ry.”

(Nec

hes,

Fik

es, F

inin

, Gru

ber,

Sen

ator

, Sw

arto

ut, 1

991)

Exa

mpl

eG

EN

E O

NT

OL

OG

Y (

GO

)

imm

une

resp

onse

i-

acut

e-ph

ase

resp

onse

i-

anap

hyla

xis

i-an

tige

n pr

esen

tatio

n i-

anti

gen

proc

essi

ngi-

cell

ular

def

ense

res

pons

ei-

cyto

kine

met

abol

ism

i-

cyto

kine

bio

synt

hesi

s sy

nony

mcy

toki

ne p

rodu

ctio

n…

p-re

gula

tion

of c

ytok

ine

bios

ynth

esis

…… i-

B-c

ell a

ctiv

atio

n

i-B

-cel

l dif

fere

ntia

tion

i-

B-c

ell p

roli

fera

tion

i-

cellu

lar

defe

nse

resp

onse

… i-

T-c

ell a

ctiv

atio

n

i-ac

tiva

tion

of

natu

ral k

ille

r ce

ll a

ctiv

ity

Ont

olog

ies

used

nfo

r co

mm

unic

atio

n be

twee

n pe

ople

and

or

gani

zatio

nsn

for

enab

ling

know

ledg

e re

use

and

shar

ing

nas

bas

is f

or in

tero

pera

bilit

y be

twee

n sy

stem

sn

as r

epos

itory

of

info

rmat

ion

nas

que

ry m

odel

for

info

rmat

ion

sour

ces

Key

tech

nolo

gy f

or th

e Se

man

tic W

eb

Bio

med

ical

Ont

olog

ies

-ef

fort

s

OB

O –

Ope

n B

iom

edic

al O

ntol

ogie

sht

tp://

ww

w.o

bofo

undr

y.or

g/(o

ver

50 o

ntol

ogie

s)

”T

he m

issi

on o

f O

BO

is to

sup

port

com

mun

ity

mem

bers

who

are

de

velo

ping

and

publ

ishi

ngon

tolo

gies

in th

e bi

omed

ical

dom

ain.

It i

s ou

rvi

sion

that

a c

ore

of th

ese

onto

logi

esw

illbe

fu

llyin

tero

pera

ble,

by

virt

ueof

a c

omm

onde

sign

phi

loso

phy

and

impl

emen

tatio

n, th

ereb

yen

ablin

gsc

ient

ists

and

thei

rin

stru

men

ts to

com

mun

icat

ew

ith

min

imum

am

bigu

ity.

In

this

way

the

data

gen

erat

edin

the

cour

seof

bio

med

ical

rese

arch

will

form

a s

ingl

e, c

onsi

sten

t, cu

mul

ativ

ely

expa

ndin

g, a

nd a

lgor

ithm

ical

lytr

acta

ble

who

le. T

his

core

will

be k

now

nas

the

"OB

O F

ound

ry".

.”

OB

O F

ound

ry

1.op

enan

d av

aila

ble

2.co

mm

onsh

ared

synt

ax

3.un

ique

iden

tifie

rsp

ace

4.pr

oced

ures

for

iden

tifyi

ngdi

stin

ctsu

cces

sive

ver

sion

s5.

clea

rly

spec

ifie

dan

d cl

earl

yde

linea

ted

cont

ent

6.te

xtua

ldef

initi

ons

for

all t

erm

s7.

use

rela

tions

fro

m O

BO

Rel

atio

n O

ntol

ogy

8.w

elld

ocum

ente

d9.

plur

ality

of in

depe

nden

t use

rs10

.de

velo

ped

colla

bora

tivel

yw

ith o

ther

OB

O F

ound

rym

embe

rs

Bio

med

ical

Ont

olog

ies

-ef

fort

s

Nat

iona

l Cen

ter

for

Bio

med

ical

Ont

olog

y ht

tp://

bioo

ntol

ogy.

org/

inde

x.ht

ml

Fund

edby

Nat

iona

l Ins

titut

esof

Hea

lth

”The

goa

lof

the

Cen

ter

is to

sup

port

bio

med

ical

rese

arch

ers

in

thei

rkn

owle

dge-

inte

nsiv

ew

ork,

by

prov

idin

gon

line

tool

san

d a

Web

por

tal e

nabl

ing

them

to a

cces

s, r

evie

w, a

nd in

tegr

ate

disp

arat

e on

tolo

gica

lres

ourc

esin

all

aspe

cts

of b

iom

edic

alin

vest

igat

ion

and

clin

ical

prac

tice.

A m

ajor

foc

usof

our

wor

kin

volv

esth

e us

eof

bio

med

ical

onto

logi

esto

aid

in th

e m

anag

emen

t and

ana

lysi

sof

dat

a de

rive

dfr

om c

ompl

exex

peri

men

ts.”

Syst

ems

Bio

logy

Ont

olog

ies

-ef

fort

s

nSy

stem

s B

iolo

gy O

ntol

ogy

nPr

oteo

mic

sSt

anda

rd I

nitia

tive

for

Mol

ecul

arIn

tera

ctio

n

nB

ioPA

X

Ont

olog

yA

lignm

ent

nnO

ntol

ogy

alig

nmen

tO

ntol

ogy

alig

nmen

t

nO

ntol

ogy

alig

nmen

t str

ateg

ies

nE

valu

atio

n of

ont

olog

y al

ignm

ent s

trat

egie

s

nC

urre

ntis

sues

nO

ntol

ogy-

base

dlit

erat

ure

sear

ch

Ont

olog

ies

in b

iom

edic

al r

esea

rch

nm

any

biom

edic

al o

ntol

ogie

s

npr

actic

al u

se o

f bi

omed

ical

onto

logi

ese.

g. d

atab

ases

ann

otat

ed w

ith

GO

GE

NE

ON

TO

LO

GY

(G

O)

imm

une

resp

onse

i-

acut

e-ph

ase

resp

onse

i-

anap

hyla

xis

i-an

tigen

pre

sent

atio

n i-

antig

en p

roce

ssin

gi-

cellu

lar

defe

nse

resp

onse

i-cy

toki

ne m

etab

olis

m

i-cy

toki

ne b

iosy

nthe

sis

syno

nym

cyto

kine

pro

duct

ion

…p-

regu

latio

n of

cyt

okin

e bi

osyn

thes

is…

… i-B

-cel

l act

ivat

ion

i-

B-c

ell d

iffe

rent

iatio

n i-

B-c

ell p

rolif

erat

ion

i-

cellu

lar

defe

nse

resp

onse

… i-

T-c

ell a

ctiv

atio

n

i-ac

tivat

ion

of n

atur

al k

iller

ce

ll ac

tivit

y …

Ont

olog

ies

with

ove

rlap

ping

in

form

atio

n

SIG

NA

L-O

NT

OL

OG

Y (

SigO

)

Imm

une

Res

pons

ei-

Alle

rgic

Res

pons

ei-

Ant

igen

Pro

cess

ing

and

Pre

sent

atio

ni-

B C

ell A

ctiv

atio

ni-

B C

ell D

evel

opm

ent

i-C

ompl

emen

t Sig

nalin

g sy

nony

m c

ompl

emen

t act

ivat

ion

i-C

ytok

ine

Res

pons

e i-

Imm

une

Supp

ress

ion

i-In

flam

mat

ion

i-In

test

inal

Im

mun

ity

i-L

euko

trie

ne R

espo

nse

i-L

euko

trie

ne M

etab

olis

m

i-N

atur

al K

iller

Cel

l Res

pons

ei-

T C

ell A

ctiv

atio

ni-

T C

ell D

evel

opm

ent

i-T

Cel

l Sel

ectio

n in

Thy

mus

GE

NE

ON

TO

LO

GY

(G

O)

imm

une

resp

onse

i-ac

ute-

phas

e re

spon

se

i-an

aphy

laxi

s i-

antig

en p

rese

ntat

ion

i-an

tigen

pro

cess

ing

i-ce

llula

r de

fens

e re

spon

sei-

cyto

kine

met

abol

ism

i-

cyto

kine

bio

synt

hesi

ssy

nony

m c

ytok

ine

prod

uctio

n…

p-re

gula

tion

of c

ytok

ine

bios

ynth

esis

…… i-

B-c

ell a

ctiv

atio

ni-

B-c

ell d

iffe

rent

iatio

n i-

B-c

ell p

rolif

erat

ion

i-

cellu

lar

defe

nse

resp

onse

… i-

T-c

ell a

ctiv

atio

ni-

activ

atio

n of

nat

ural

kill

er

cell

activ

ity

Ont

olog

ies

with

ove

rlap

ping

in

form

atio

nn

Use

of

mul

tiple

ont

olog

ies

e.g.

cus

tom

-spe

cifi

c on

tolo

gy +

sta

ndar

d on

tolo

gydi

ffer

ent v

iew

s on

sam

e do

mai

nco

nnec

ting

rela

ted

area

s

nB

otto

m-u

p cr

eatio

n of

ont

olog

ies

expe

rts

can

focu

s on

thei

r do

mai

n of

exp

ertis

e

impo

rtan

t to

know

the

inte

rim

port

ant t

o kn

ow th

e in

ter --

onto

logy

on

tolo

gy

rela

tion

ship

sre

lati

onsh

ips

SIG

NA

L-O

NT

OL

OG

Y (

SigO

)

Imm

une

Res

pons

ei-

Alle

rgic

Res

pons

ei-

Ant

igen

Pro

cess

ing

and

Pre

sent

atio

ni-

B C

ell A

ctiv

atio

n i-

B C

ell D

evel

opm

ent

i-C

ompl

emen

t Sig

nalin

g sy

nony

m c

ompl

emen

t act

ivat

ion

i-C

ytok

ine

Res

pons

e i-

Imm

une

Supp

ress

ion

i-In

flam

mat

ion

i-In

test

inal

Im

mun

ity

i-L

euko

trie

ne R

espo

nse

i-L

euko

trie

ne M

etab

olis

m

i-N

atur

al K

iller

Cel

l Res

pons

e i-

T C

ell A

ctiv

atio

n i-

T C

ell D

evel

opm

ent

i-T

Cel

l Sel

ectio

n in

Thy

mus

GE

NE

ON

TO

LO

GY

(G

O)

imm

une

resp

onse

i-

acut

e-ph

ase

resp

onse

i-

anap

hyla

xis

i-an

tigen

pre

sent

atio

n i-

antig

en p

roce

ssin

gi-

cellu

lar

defe

nse

resp

onse

i-cy

toki

ne m

etab

olis

m

i-cy

toki

ne b

iosy

nthe

sis

syno

nym

cyt

okin

e pr

oduc

tion

…p-

regu

latio

n of

cyt

okin

e bi

osyn

thes

is…

… i-B

-cel

l act

ivat

ion

i-

B-c

ell d

iffe

rent

iatio

n i-

B-c

ell p

rolif

erat

ion

i-

cellu

lar

defe

nse

resp

onse

… i-

T-c

ell a

ctiv

atio

n

i-ac

tivat

ion

of n

atur

al k

iller

ce

ll ac

tivit

y…

Ont

olog

y A

lignm

ent

equi

vale

nt c

once

pts

equi

vale

nt r

elat

ions

is-a

rel

atio

n

SIG

NA

L-O

NT

OL

OG

Y (

SigO

)

Imm

une

Res

pons

ei-

Alle

rgic

Res

pons

ei-

Ant

igen

Pro

cess

ing

and

Pre

sent

atio

ni-

B C

ell A

ctiv

atio

ni-

B C

ell D

evel

opm

ent

i-C

ompl

emen

t Sig

nalin

g sy

nony

m c

ompl

emen

t act

ivat

ion

i-C

ytok

ine

Res

pons

e i-

Imm

une

Supp

ress

ion

i-In

flam

mat

ion

i-In

test

inal

Im

mun

ity

i-L

euko

trie

ne R

espo

nse

i-L

euko

trie

ne M

etab

olis

m

i-N

atur

al K

iller

Cel

l Res

pons

ei-

T C

ell A

ctiv

atio

ni-

T C

ell D

evel

opm

ent

i-T

Cel

l Sel

ectio

n in

Thy

mus

GE

NE

ON

TO

LO

GY

(G

O)

imm

une

resp

onse

i-ac

ute-

phas

e re

spon

se

i-an

aphy

laxi

s i-

antig

en p

rese

ntat

ion

i-an

tigen

pro

cess

ing

i-ce

llula

r de

fens

e re

spon

sei-

cyto

kine

met

abol

ism

i-

cyto

kine

bio

synt

hesi

ssy

nony

m c

ytok

ine

prod

uctio

n…

p-re

gula

tion

of c

ytok

ine

bios

ynth

esis

…… i-

B-c

ell a

ctiv

atio

ni-

B-c

ell d

iffe

rent

iatio

n i-

B-c

ell p

rolif

erat

ion

i-

cellu

lar

defe

nse

resp

onse

… i-

T-c

ell a

ctiv

atio

ni-

activ

atio

n of

nat

ural

kill

er

cell

activ

ity

Def

inin

g th

e re

latio

ns b

etw

een

the

term

s in

dif

fere

nt o

ntol

ogie

s

Ont

olog

yA

lignm

ent

nO

ntol

ogy

alig

nmen

t

nnO

ntol

ogy

alig

nmen

t str

ateg

ies

Ont

olog

y al

ignm

ent s

trat

egie

s

nE

valu

atio

n of

ont

olog

y al

ignm

ent s

trat

egie

s

nC

urre

ntis

sues

nO

ntol

ogy-

base

dlit

erat

ure

sear

ch

An

Ali

gnm

entF

ram

ewor

k

Pre

proc

essi

ng

Prep

roce

ssin

g

For

exam

ple,

nSe

lect

ion

of f

eatu

res

nSe

lect

ion

of s

earc

h sp

ace

Mat

cher

s

nSt

rate

gies

bas

ed o

n lin

guis

tic m

atch

ing

nSt

ruct

ure-

base

d st

rate

gies

nC

onst

rain

t-ba

sed

appr

oach

es

nIn

stan

ce-b

ased

stra

tegi

es

nU

seof

aux

iliar

yin

form

atio

n

Mat

cher

Str

ateg

ies

nnSt

rate

gies

bas

ed o

n lin

guis

tic m

atch

ing

Stra

tegi

es b

ased

on

lingu

istic

mat

chin

g

SigO

: c

ompl

emen

t si

gnal

ing

syno

nym

com

plem

ent

acti

vati

on

GO

:C

ompl

emen

t A

ctiv

atio

n

Exa

mpl

em

atch

ers

nE

dit d

ista

nce

¤N

umbe

rof

del

etio

ns, i

nser

tions

, sub

stitu

tions

req

uire

dto

tran

sfor

m o

nest

ring

into

anot

her

¤aa

aaba

ab: e

ditd

ista

nce

2

nN

-gra

N-g

ram

: N

con

secu

tive

char

acte

rsin

a s

trin

g

¤Si

mila

rity

base

don

set

com

pari

son

of n

-gra

ms

¤aa

aa: {

aa, a

a, a

a};

baa

b: {

ba, a

a, a

b}

Mat

cher

Str

ateg

ies

nSt

rate

gies

bas

ed o

n lin

guis

tic m

atch

ing

nnSt

ruct

ure

Stru

ctur

e --ba

sed

stra

tegi

esba

sed

stra

tegi

es

nC

onst

rain

t-ba

sed

appr

oach

es

nIn

stan

ce-b

ased

stra

tegi

es

nU

seof

aux

iliar

yin

form

atio

n

Exa

mpl

em

atch

ers

nPr

opag

atio

nof

sim

ilari

tyva

lues

nA

ncho

red

mat

chin

g

Exa

mpl

em

atch

ers

nPr

opag

atio

nof

sim

ilari

tyva

lues

nA

ncho

red

mat

chin

g

Exa

mpl

em

atch

ers

nPr

opag

atio

nof

sim

ilari

tyva

lues

nA

ncho

red

mat

chin

g

Mat

cher

Str

ateg

ies

nSt

rate

gies

bas

ed o

n lin

guis

tic m

atch

ing

nSt

ruct

ure-

base

d st

rate

gies

nnC

onst

rain

tC

onst

rain

t --ba

sed

base

dap

proa

ches

appr

oach

es

nIn

stan

ce-b

ased

stra

tegi

es

nU

seof

aux

iliar

yin

form

atio

n

O1

O2

Bir

d

Mam

mal

Mam

mal

Fly

ing

Ani

mal

Mat

cher

Str

ateg

ies

nSt

rate

gies

bas

ed o

n lin

guis

tic m

atch

ing

nSt

ruct

ure-

base

d st

rate

gies

nnC

onst

rain

tC

onst

rain

t --ba

sed

base

dap

proa

ches

appr

oach

es

nIn

stan

ce-b

ased

stra

tegi

es

nU

seof

aux

iliar

yin

form

atio

n

O1

O2

Bir

d

Mam

mal

Mam

mal

Ston

e

Exa

mpl

em

atch

ers

nSi

mila

ritie

sbe

twee

nda

ta ty

pes

nSi

mila

ritie

sba

sed

on c

ardi

nalit

ies

Mat

cher

Str

ateg

ies

nSt

rate

gies

bas

ed o

n lin

guis

tic m

atch

ing

nSt

ruct

ure-

base

d st

rate

gies

nC

onst

rain

t-ba

sed

appr

oach

es

nnIn

stan

ceIn

stan

ce-- b

ased

base

dst

rate

gies

stra

tegi

es

nU

seof

aux

iliar

yin

form

atio

n

Ont

olog

y

inst

ance

corp

us

Exa

mpl

em

atch

ers

nIn

stan

ce-b

ased

nU

selif

e sc

ienc

e lit

erat

ure

as in

stan

ces

Lea

rnin

g m

atch

ers

–in

stan

ce-b

ased

st

rate

gies

nB

asic

intu

ition

A

sim

ilari

tym

easu

rebe

twee

nco

ncep

tsca

nbe

co

mpu

ted

base

don

the

prob

abili

tyth

at

docu

men

tsab

outo

neco

ncep

tare

als

oab

outt

he

othe

rco

ncep

tand

vic

e ve

rsa.

Bas

ic N

aïve

Bay

esm

atch

er

nG

ener

ate

corp

ora

¤U

seco

ncep

tas

quer

yte

rm in

Pub

Med

¤R

etri

eve

mos

trec

ent P

ubM

edab

stra

cts

nG

ener

ate

clas

sifi

ers

¤N

aive

Bay

escl

assi

fier

s, o

nepe

r on

tolo

gy

nC

lass

ific

atio

Abs

trac

ts r

elat

edto

one

onto

logy

are

clas

sifi

edto

the

conc

ept

in th

e ot

her

onto

logy

with

hig

hest

post

erio

rpr

obab

ility

P(C

|d)

nC

alcu

late

sim

ilar

itie

s

Mat

cher

Str

ateg

ies

nSt

rate

gies

bas

ed li

ngui

stic

mat

chin

g

nSt

ruct

ure-

base

d st

rate

gies

nC

onst

rain

t-ba

sed

appr

oach

es

nIn

stan

ce-b

ased

stra

tegi

es

nnU

seU

seof

of

aux

iliar

yau

xilia

ryin

form

atio

nin

form

atio

nthes

auri al

ignm

ent

stra

tegi

es

dict

iona

ry

inte

rmed

iate

onto

logy

Exa

mpl

em

atch

ers

nU

seof

Wor

dNet

¤U

seW

ordN

etto

fin

dsy

nony

ms

¤U

seW

ordN

etto

fin

dan

cest

ors

and

desc

enda

nts

in th

e is

-a

hier

arch

y

nU

seof

Uni

fied

Med

ical

Lan

guag

e Sy

stem

(U

ML

S)¤

Incl

udes

man

yon

tolo

gies

¤In

clud

esm

any

map

ping

s(n

ot c

ompl

ete)

¤U

seU

ML

S m

appi

ngs

in th

e co

mpu

tatio

nof

the

sim

ilar

ity

valu

es

Ontology Alignment and Mergning Systems

Com

bina

tion

s

Com

bina

tion

Stra

tegi

es

nU

sual

ly w

eigh

ted

sum

of

sim

ilari

ty v

alue

s of

di

ffer

ent m

atch

ers

nM

axim

um o

f si

mila

rity

val

ues

of d

iffe

rent

m

atch

ers

Filt

erin

g

nT

hres

hold

filte

ring

Pair

s of

con

cept

s w

ith s

imila

rity

hig

her

or e

qual

th

an th

resh

old

are

map

ping

sug

gest

ions

Filte

ring

tech

niqu

es

th

( 2,

B )

( 3,

F )

( 6,

D )

( 4,

C )

( 5,

C )

( 5,

E )

……

sugg

est

disc

ard

sim

Filte

ring

tech

niqu

es

low

er-t

h

( 2,

B )

( 3,

F )

( 6,

D )

( 4,

C )

( 5,

C )

( 5,

E )

……

uppe

r-th

nD

oubl

eth

resh

old

filte

ring

(1)

Pair

s of

con

cept

s w

ith s

imila

rity

hig

her

than

or

equa

l to

uppe

rth

resh

old

are

map

ping

sug

gest

ions

(2)

Pair

s of

con

cept

s w

ith s

imila

rity

bet

wee

n lo

wer

and

uppe

rth

resh

olds

are

m

appi

ng s

ugge

stio

ns if

they

mak

e se

nse

with

res

pect

to th

e st

ruct

ure

of th

e on

tolo

gies

and

the

sugg

esti

ons

acco

rdin

g to

(1)

Exa

mpl

e al

ignm

ent s

yste

m S

AM

BO

–pr

epro

cess

ing,

mat

cher

s, c

ombi

nati

on, f

ilter

Exa

mpl

e al

ignm

ent s

yste

m S

AM

BO

–su

gges

tion

mod

e

Exa

mpl

e al

ignm

ent s

yste

m

SAM

BO

–m

anua

l mod

e

Ont

olog

yA

lignm

ent

nO

ntol

ogy

alig

nmen

t

nO

ntol

ogy

alig

nmen

t str

ateg

ies

nnE

valu

atio

n of

ont

olog

y al

ignm

ent s

trat

egie

s E

valu

atio

n of

ont

olog

y al

ignm

ent s

trat

egie

s

nC

urre

ntis

sues

nO

ntol

ogy-

base

dlit

erat

ure

sear

ch

Eva

luat

ion

mea

sure

s

nPr

ecis

ion:

#

corr

ect s

ugge

sted

map

ping

s #

sugg

este

d m

appi

ngs

nR

ecal

l: #

corr

ect s

ugge

sted

map

ping

s #

corr

ect m

appi

ngs

nF-

mea

sure

: com

bina

tion

of p

reci

sion

and

re

call

Ont

olog

yA

lignm

ent

Eva

luat

ion

Init

iati

ve

OA

EI

nSi

nce

2004

nE

valu

atio

n of

sys

tem

s

nD

iffe

rent

trac

ks¤

com

pari

son:

ben

chm

ark

(ope

n)

¤ex

pres

sive

: ana

tom

y (b

lind)

, fis

heri

es (

expe

rt)

¤di

rect

orie

s an

d th

esau

ri: d

irec

tory

, lib

rary

, cr

ossl

ingu

alre

sour

ces

(blin

d)

¤co

nsen

sus:

con

fere

nce

OA

EI

2007

n17

sys

tem

s pa

rtic

ipat

ed¤

benc

hmar

k (1

3)n

ASM

OV

: p =

0.9

5, r

= 0

.90

¤an

atom

y (1

1)

nA

OA

S: f

= 0

.86,

r+

= 0

.50

nSA

MB

O: f

=0.

81, r

+ =

0.5

8

¤lib

rary

(3)

nT

hesa

urus

mer

ging

: FA

LC

ON

: p =

0.9

7, r

= 0

.87

nA

nnot

atio

n sc

enar

io:

¤FA

LC

ON

: pb

=0.

65, r

b=

0.49

, pa

= 0

.52,

ra

= 0.

36, J

a=

0.30

¤Si

las:

pb

= 0

.66,

rb=

0.4

7, p

a =

0.53

, ra

= 0

.35,

Ja

= 0

.29

¤di

rect

ory

(9),

foo

d (6

), e

nvir

onm

ent (

2), c

onfe

renc

e (6

)

OA

EI

2008

–an

atom

y tr

ack

nA

lign

¤M

ouse

ana

tom

y: 2

744

term

NC

I-an

atom

y: 3

304

term

Map

ping

s: 1

544

(of

whi

ch 9

34 ‘

triv

ial’

)

nT

asks

¤

1. A

lign

and

optim

ize

2-3.

Alig

n an

d op

timiz

e p

/ r¤

4. A

lign

whe

n pa

rtia

l ref

eren

ce a

lignm

ent i

s gi

ven

and

optim

ize

f

OA

EI

2008

–an

atom

y tr

ack#

1

n9

syst

ems

part

icip

ated

nSA

MB

p=0.

869,

r=

0.83

6, r

+=

0.58

6, f

=0.

852

nSA

MB

Odt

p=0.

831,

r=

0.83

3, r

+=

0.57

9, f

=0.

832

nU

se o

f T

erm

WN

and

UM

LS

OA

EI

2008

–an

atom

y tr

ack#

1

Is b

ackg

roun

d kn

owle

dge

(BK

) ne

eded

?

Of

the

non-

triv

ial m

appi

ngs:

¤C

a 50

% f

ound

by

syst

ems

usin

g B

K a

nd s

yste

ms

not

usin

g B

Ca

13%

fou

nd o

nly

by s

yste

ms

usin

g B

Ca

13%

fou

nd o

nly

by s

yste

ms

not u

sing

BK

¤C

a 25

% n

ot f

ound

Proc

essi

ng ti

me:

ho

urs

with

BK

, min

utes

with

out B

K

OA

EI

2008

–an

atom

y tr

ack#

4

Can

we

use

give

n m

appi

ngs

whe

n co

mpu

ting

sugg

estio

ns?

part

ial r

efer

ence

alig

nmen

t giv

en w

ith a

ll tr

ivia

l and

50

non-

triv

ial m

appi

ngs

nSA

MB

p=0.

636

0.66

0, r

=0.

626

0.62

4, f

=0.

631

0.64

2

nSA

MB

Odt

p=0.

563

0.60

3, r

=0.

622

0.63

0, f

=0.

591

0.61

6

(mea

sure

s co

mpu

ted

on n

on-g

iven

par

t of

the

refe

renc

e al

ignm

ent)

OA

EI

2007

-200

8

nSy

stem

s ca

n us

e on

ly o

ne c

ombi

natio

n of

st

rate

gies

per

task

syst

ems

use

sim

ilar

stra

tegi

es¤

text

: str

ing

mat

chin

g, tf

-idf

¤st

ruct

ure:

pro

paga

tion

of s

imil

arity

to a

nces

tors

an

d/or

des

cend

ants

¤th

esau

rus

(Wor

dNet

)

¤do

mai

n kn

owle

dge

impo

rtan

t for

ana

tom

y ta

sk?

Ont

olog

yA

lignm

ent

nO

ntol

ogy

alig

nmen

t

nO

ntol

ogy

alig

nmen

t str

ateg

ies

nE

valu

atio

n of

ont

olog

y al

ignm

ent s

trat

egie

s

nnC

urre

nt I

ssue

sC

urre

nt I

ssue

s

nO

ntol

ogy-

base

dlit

erat

ure

sear

ch

Cur

rent

issu

es

nSy

stem

s an

d al

gori

thm

Com

plex

onto

logi

es

¤U

seof

inst

ance

-bas

edte

chni

ques

¤A

lignm

entt

ypes

(equ

ival

ence

, is-

a, …

)

¤C

ompl

exm

appi

ngs

(1-n

, m-n

)

¤C

onne

ctio

n on

tolo

gyty

pes

–al

ignm

ents

trat

egie

s

nE

valu

atio

SEA

LS

–S

eman

ticE

valu

atio

nA

t Lar

geSc

ale

Cur

rent

issu

es

nR

ecom

men

ding

’bes

t’al

ignm

ents

trat

egie

s

nU

seof

Par

tialR

efer

ence

Alig

nmen

t

----

----

----

----

----

----

----

----

----

----

----

----

----

----

-

nIn

tegr

atio

n of

ont

olog

yal

ignm

enta

nd r

epai

rof

th

e st

ruct

ure

of o

ntol

ogie

s

Ont

olog

yA

lignm

ent

nO

ntol

ogy

alig

nmen

t

nO

ntol

ogy

alig

nmen

t str

ateg

ies

nE

valu

atio

n of

ont

olog

y al

ignm

ent s

trat

egie

s

nC

urre

nt is

sues

nnO

ntol

ogy

Ont

olog

y --ba

sed

liter

atur

e se

arch

base

d lit

erat

ure

sear

ch

Lit

erat

ure

sear

ch

nH

uge

amou

nt o

f sc

ient

ific

lite

ratu

re.

nN

eed

to in

tegr

ate

a sp

ectr

um o

f in

form

atio

n to

pe

rfor

m a

task

.

Lit

erat

ure

sear

ch

nH

ow to

kno

w w

hat i

s in

the

repo

sito

ry¤

Lac

k of

kno

wle

dge

of th

e do

mai

n

nH

ow to

com

pose

an

expr

essi

ve q

uery

¤L

ack

of k

now

ledg

e of

sea

rch

tech

nolo

gy

Exa

mpl

e sc

enar

io“L

ipid

nK

eyw

ord

sear

ch r

etur

ns a

ll do

cum

ents

co

ntai

ning

lipi

d.¤

No

know

ledg

e; te

rmin

olog

y pr

oble

m

nR

elat

ions

hips

: use

of

mul

tiple

key

wor

ds

with

/with

out b

oole

anop

erat

ors,

e.g.

lipi

d an

d di

seas

e

Exa

mpl

e sc

enar

io“L

ipid

nK

eyw

ord

sear

ch r

etur

ns a

list

of

rele

vant

qu

esti

ons

conc

erni

ng li

pid.

Use

r se

lect

s qu

esti

on

and

retr

ieve

s kn

owle

dge

and

prov

enan

ce

docu

men

ts.

nM

ultip

le s

earc

h te

rms:

req

uire

men

t tha

t the

re a

re

rele

vant

con

nect

ions

bet

wee

n th

e ke

ywor

ds.

lipid

Rel

evan

t qu

erie

s

nR

elev

ant q

uery

incl

udin

ga

num

ber

of c

once

pts

and

rela

tions

fro

m a

n on

tolo

gy

conn

ecte

dsu

b-gr

aph

of th

e on

tolo

gyth

at in

clud

esth

e co

ncep

tsan

d re

latio

ns.

(que

rygr

aph

base

don

the

conc

epts

and

rela

tion

s;

slic

e is

set

of a

ll q

uery

grap

hsba

sed

on th

e co

ncep

tsan

d re

lati

ons)

Que

ry g

raph

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

1

4

67

23

5

e3

e4e5

e6

e7

e1e2

Que

ry g

raph

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

1

4

67

23

5

e3

e4e5

e6

e7

e1e2

Que

ry g

raph

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

����

1

4

67

23

5

e3

e4e5

e6

e7

e1e2

Spec

ial c

ases

nN

o re

latio

ns, s

ever

alco

ncep

ts¤

Rel

evan

t que

ries

rega

rdin

gco

ncep

ts; r

elat

ions

are

su

gges

ted

by th

e sy

stem

.

¤D

iffe

renc

ew

ith tr

aditi

onal

tech

niqu

es: e

xtra

req

uire

men

tth

at s

earc

hte

rms

need

to b

e co

nnec

ted

in th

e on

tolo

gy.

nN

o re

latio

ns, o

neco

ncep

Rel

evan

t que

ries

incl

udin

ga

spec

ific

quer

yte

rm.

¤C

ompu

tes

the

onto

logi

cale

nvir

onm

ento

f th

e qu

ery

term

.

Rel

evan

t qu

erie

s–

mul

tipl

e on

tolo

gies

nR

elev

ant q

uery

incl

udin

ga

num

ber

of c

once

pts

and

rela

tions

fro

m m

ultip

le o

ntol

ogie

s

Que

ry g

raph

sco

nnec

ted

by a

pat

hgo

ing

thro

ugh

a m

appi

ngin

the

alig

nmen

t. (a

lign

edqu

ery

grap

hba

sed

on q

uery

grap

hs;

alig

ned

slic

e is

set

of a

ll a

lign

edqu

ery

grap

hsba

sed

on th

e qu

ery

grap

hs)

Alig

ned

quer

ygr

aph

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

1

4

67

23

5

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

A

C

DBE F

e11

e12

e13

e14

e15

ea1

ea2

e21

e22

e23

e24

e25

e26

e16

e17

Alig

ned

quer

ygr

aph

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

1

4

67

23

5

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

A

C

DBE F

e11

e12

e13

e14

e15

ea1

ea2

e21

e22

e23

e24

e25

e26

e16

e17

Alig

ned

quer

ygr

aph

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

1

4

67

23

5

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

A

C

DBE F

e11

e12

e13

e14

e15

ea1

ea2

e21

e22

e23

e24

e25

e26

e16

e17

Fra

mew

ork

Ext

erna

lres

ourc

es

nL

itera

ture

docu

men

tbas

Gen

erat

edfr

om a

col

lect

ion

of 7

498

PubM

edab

stra

cts

rele

vant

for

Ova

rian

Can

cer.

683

pap

ers

incl

uded

lipid

nam

esfr

om w

hich

241

full

pape

rsw

ere

dow

nloa

dabl

e.

nO

ntol

ogy

and

onto

logy

alig

nmen

trep

osito

ry¤

Lip

id o

ntol

ogy

¤Si

gnal

ont

olog

Alig

men

t usi

ngSA

MB

O

2) S

ente

nce

Ext

ract

ion

1) D

ocum

ent C

onte

nt

3) S

ente

nce

Det

ectio

n: li

pid

inte

ract

ion

prot

ein

4) E

ntity

Rec

ogni

tion:

te

rm id

entif

icat

ion

/ ass

ign

lipid

clas

s

5) N

orm

aliz

atio

n: c

olla

pse

lipid

syno

nym

s

6) R

elat

ion

Ext

ract

ion:

Lip

id-P

rote

in o

r Li

pid

Dis

ease

8) P

opul

ate

OW

L on

tolo

gy (

JEN

A -

AP

I)

Com

plet

e In

stan

tiate

d O

WL-

DL

Ont

olog

y

Ter

m L

ist D

B’s

:Li

pid

nam

es,

LIP

IDM

AP

S, L

ipid

Ban

k,

KE

GG

cla

ssifi

catio

ns,

Dis

ease

nam

es,

Pro

tein

nam

esS

tem

med

Inte

ract

ions

Doc

umen

t and

se

nten

ce m

eta

data

"T

LR4

bind

s to

PO

PC

", ta

gged

as

"TLR

4 bi

nds

to P

OP

C",

tagg

ed a

s "<

term

cat

egor

y="

"<te

rm c

ateg

ory=

" pro

tein

prot

ein

"> T

LR4<

/term

>

"> T

LR4<

/term

>

bind

s to

bi

nds

to

<te

rm c

ateg

ory=

"<

term

cat

egor

y=" l

ipid

lipid

">P

OP

C<

/term

>"

">P

OP

C<

/term

>"

7) C

lass

ifica

tion:

Iden

tify

onto

logy

cla

sses

and

spe

cify

rela

tions

for

all s

ente

nces

, pro

tein

s,lip

idsu

bcla

sses

.

Kno

wle

dge

base

inst

anti

atio

n

Lipid Instance

Lipid Instance

Lipid Class

Protein

Instance

Kno

wle

dge

base

inst

anti

atio

n

Slic

e ge

nera

tion

nC

urre

ntim

plem

enta

tion

focu

ses

on s

lices

ba

sed

on c

once

pts.

nD

epth

-fir

sttr

aver

salo

f on

tolo

gyto

fin

dpa

ths

betw

een

give

n co

ncep

ts; p

aths

can

be p

utto

geth

erto

fin

dsl

ices

/que

rygr

aphs

.

Slic

e al

ignm

ent

nA

lgor

ithm

com

pute

ssu

bset

of a

ligne

dsl

ice.

nA

ssum

ptio

n: s

hort

erpa

ths

repr

esen

tclo

ser

rela

tion

ship

s.

nA

lgor

ithm

conn

ects

slic

es u

sing

shor

test

path

sfr

om g

iven

con

cept

sin

one

onto

logy

to g

iven

co

ncep

tsin

oth

eron

tolo

gy.

Slic

ing

thro

ugh

the

liter

atur

e

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

������

������

������

������

������

������

������

������

������

������

������

������

������

������

������

������

1

4

67

23

5

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

�����

A

C

DBE F

e11

e12

e13

e14

e15

ea1

ea2

e21

e22

e23

e24

e25

e26

e16

e17

prot

ein

lipid

dise

ase

Sig

nal-p

athw

ay Invo

lved

-inIn

tera

cts-

with

Impl

icat

ed-in

Nat

ural

lang

uage

quer

yge

nera

tion

nT

ripl

ere

pres

enta

tion:

<li

pid,

inte

ract

s-w

ith,

pro

tein

>

nR

ule

base

to g

ener

ate

NL

sta

tem

ents

.

Wha

tlip

id in

tera

cts

wit

h pr

otei

ns?

¤L

earn

edfr

om e

xam

ples

.

nA

ggre

gatio

n of

sta

tem

ents

from

dif

fere

nt

trip

les,

gra

mm

arch

ecki

ng.

Que

ry

nSe

ndnR

QL

quer

yto

RA

CE

R.

Fut

ure

Wor

k

nT

rade

off

in q

uery

gen

erat

ion

betw

een

com

plet

enes

s an

d in

form

atio

n ov

erlo

ad.

nR

elev

ance

mea

sure

and

que

ry r

anki

ng.

nIn

tegr

ated

impl

emen

tatio

n.

nSc

alab

ility

test

ing.

Furt

her

read

ing

Ont

olog

y al

ignm

ent

-ge

nera

l

nht

tp://

ww

w.o

ntol

ogym

atch

ing.

org

(ple

nty

of r

efer

ence

sto

art

icle

san

d sy

stem

s)

nO

ntol

ogy

alig

nmen

t eva

luat

ion

initi

ativ

e: h

ttp://

oaei

.ont

olog

ymat

chin

g.or

g(h

ome

page

of

the

initi

ativ

e)

nE

uzen

at, S

hvai

ko, O

ntol

ogy

Mat

chin

g, S

prin

ger,

200

7.

nL

ambr

ix, S

tröm

bäck

, Tan

, Inf

orm

atio

n in

tegr

atio

n in

bio

info

rmat

ics

with

on

tolo

gies

and

stan

dard

s, in

Bry

, Mal

uszy

nski

(eds

), S

eman

tic

Tec

hniq

ues

for

the

Web

: T

he R

EW

ER

SE p

ersp

ecti

ve, c

hapt

er 8

, 343

-376

, 200

9.(c

onta

ins

curr

ently

larg

esto

verv

iew

of o

ntol

ogy

alig

nmen

tsys

tem

s)

Furt

her

read

ing

Ont

olog

y al

ignm

ent

-sy

stem

sn

Lam

brix

, Tan

, SA

MB

O –

a sy

stem

for

alig

ning

and

mer

ging

bio

med

ical

on

tolo

gies

, Jou

rnal

of W

eb S

eman

tics

, 4(3

):19

6-20

6, 2

006.

(des

crip

tion

of th

e SA

MB

O to

olan

d ov

ervi

ewof

eva

luat

ions

of d

iffe

rent

m

atch

ers)

nL

ambr

ix, T

an, A

tool

for

eva

luat

ing

onto

logy

alig

nmen

t str

ateg

ies,

Jou

rnal

on

Dat

a Se

man

tics

, VII

I:18

2-20

2, 2

007.

(des

crip

tion

of th

e K

itAM

Oto

olfo

r ev

alua

ting

mat

cher

s)

Furt

her

read

ing

Ont

olog

yal

ignm

ent

-re

com

men

dati

onof

alig

nmen

tst

rate

gies

nT

an, L

ambr

ix, A

met

hod

for

reco

mm

endi

ng o

ntol

ogy

alig

nmen

t str

ateg

ies,

In

tern

atio

nal S

eman

tic

Web

Con

fere

nce,

494

-507

, 200

7.

nE

hrig

, Sta

ab, S

ure,

Boo

tstr

appi

ng o

ntol

ogy

alig

nmen

t met

hods

with

A

PFE

L, I

nter

nati

onal

Sem

anti

c W

eb C

onfe

renc

e, 1

86-2

00, 2

005.

nM

ocho

l, Je

ntzs

ch, E

uzen

at, A

pply

ing

an a

naly

ticm

etho

dfo

r m

atch

ing

appr

oach

sel

ecti

on, I

nter

nati

onal

Wor

ksho

p on

Ont

olog

yM

atch

ing,

200

6.

Ont

olog

yal

ignm

ent

-P

RA

in o

ntol

ogy

alig

nmen

tn

Lam

brix

, Liu

, Usi

ngpa

rtia

lref

eren

ceal

ignm

ents

to a

lign

onto

logi

es,

Eur

opea

n Se

man

tic

Web

Con

fere

nce,

188

-202

, 200

9.

Lit

erat

ure

sear

chn

Bak

er, L

ambr

ix, L

auri

la B

ergm

an, K

anag

asab

ai, A

ng, S

licin

gth

roug

hth

e sc

ient

ific

liter

atur

e, D

ata

Inte

grat

ion

in th

e Li

fe S

cien

ces,

127

-140

, 200

9.

DIL

S 20

107t

h In

tern

atio

nal C

onfe

renc

e on

D

ata

Inte

grat

ion

in t

he L

ife

Scie

nces

Aug

ust

25-2

7, G

othe

nbur

g, S

wed

en

pape

rsu

bmis

sion

dead

line

in A

pril


Recommended