ontology mapping needs context & approximation

44
Ontology mapping needs context & approximation Frank van Harmelen Vrije Universiteit Amsterdam

Upload: abe

Post on 14-Feb-2016

68 views

Category:

Documents


7 download

DESCRIPTION

Ontology mapping needs context & approximation. Frank van Harmelen Vrije Universiteit Amsterdam. Or: . How to make ontology-mapping less like data-base integration. and more like a social conversation. Three. Two obvious intuitions. Ontology mapping needs background knowledge. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Ontology mapping  needs  context & approximation

Ontology mapping needs

context & approximation

Frank van HarmelenVrije Universiteit Amsterdam

Page 2: Ontology mapping  needs  context & approximation

2

Or:

andmore like a social conversation

How to make ontology-mapping less like data-base integration

Page 3: Ontology mapping  needs  context & approximation

3

Two obvious intuitions

Ontology mapping needs background knowledge

Ontology mapping needs approximation

The Semantic Web needs ontology mapping

Three

Page 4: Ontology mapping  needs  context & approximation

4

Which Semantic Web?Version 1:

"Semantic Web as Web of Data" (TBL)

recipe:expose databases on the web, use RDF, integrate

meta-data from: expressing DB schema semantics

in machine interpretable waysenable integration and unexpected re-

use

Page 5: Ontology mapping  needs  context & approximation

5

Which Semantic Web?Version 2:

“Enrichment of the current Web”

recipe:Annotate, classify, index

meta-data from: automatically producing markup:

named-entity recognition, concept extraction, tagging, etc.

enable personalisation, search, browse,..

Page 6: Ontology mapping  needs  context & approximation

6

Which Semantic Web?Version 1:

“Semantic Web as Web of Data”

Version 2:“Enrichment of the current Web”

Different use-cases Different techniques Different users

data-oriented

user-oriented

Page 7: Ontology mapping  needs  context & approximation

7between source & user

Which Semantic Web?Version 1:

“Semantic Web as Web of Data”

between sources

Version 2:“Enrichment of the current Web”

But both need ontologiesfor semantic agreement

Page 8: Ontology mapping  needs  context & approximation

8

Ontology research is almost done..we know what they are

“consensual, formalised models of a domain”we know how to make and maintain

them (methods, tools, experience)

we know how to deploy them(search, personalisation, data-integration, …)

Main remaining open questions Automatic construction (learning) Automatic mapping (integration)

Page 9: Ontology mapping  needs  context & approximation

9

Three obvious intuitions

Ontology mapping needs background knowledge

Ontology mapping needs approximation

The Semantic Web needs ontology mapping

Ph.D. student

young researcher

?=?≈

AIO

post-doc

Page 10: Ontology mapping  needs  context & approximation

This work withZharko Aleksovski &

Michel Klein

Page 11: Ontology mapping  needs  context & approximation

Does context knowledge help mapping?

Page 12: Ontology mapping  needs  context & approximation

12

The general idea

source target

background knowledge

anchoring anchoring

mapping

inference

Page 13: Ontology mapping  needs  context & approximation

13

a realistic example Two Amsterdam hospitals (OLVG, AMC) Two Intensive Care Units, different vocab’s Want to compare quality of care OLVG-1400:

1400 terms in a flat list used in the first 24 hour of stay some implicit hierarchy e.g.6 types of Diabetes

Mellitus) some reduncy (spelling mistakes)

AMC: similar list, but from different hospital

Page 14: Ontology mapping  needs  context & approximation

14

Context ontology usedDICE:

2500 concepts (5000 terms), 4500 links Formalised in DL five main categories:

•tractus (e.g. nervous_system, respiratory_system)•aetiology (e.g. virus, poising)•abnormality (e.g. fracture, tumor)•action (e.g. biopsy, observation, removal)•anatomic_location (e.g. lungs, skin)

Page 15: Ontology mapping  needs  context & approximation

15

Baseline: Linguistic methods Combine lexical analysis with hierarchical

structure

313 suggested matches, around 70 % correct 209 suggested matches, around 90 % correct

High precision, low recall (“the easy cases”)

Page 16: Ontology mapping  needs  context & approximation

16

Now use background knowledge

OLVG (1400, flat)

AMC (1400, flat)

DICE(2500 concepts,

4500 links)

anchoring anchoring

mapping

inference

Page 17: Ontology mapping  needs  context & approximation

17

Example found with context knowledge (beyond lexical)

Page 18: Ontology mapping  needs  context & approximation

18

Example 2

Page 19: Ontology mapping  needs  context & approximation

19

Anchoring strengthAnchoring = substring + trivial

morphologyanchored on N aspects OLVG AMCN=5N=4N=3N=2N=1

004

144401

2198711285208

total nr. of anchored termstotal nr. of anchorings

549129

8

39% 1404

5816

96%

Page 20: Ontology mapping  needs  context & approximation

21

Experimental results Source & target =

flat lists of ±1400 ICU terms each Background = DICE (2300 concepts in DL) Manual Gold Standard (n=200)

Page 21: Ontology mapping  needs  context & approximation

Does more context knowledge help?

Page 22: Ontology mapping  needs  context & approximation

23

Adding more context Only lexical DICE (2500 concepts) MeSH (22000 concepts) ICD-10 (11000 concepts)

Anchoring strength:DICE MeSH ICD10

4 aspects 0 8 03 aspects 0 89 02 aspects 135 201 01 aspect 413 694 80total 548 992 80

Page 23: Ontology mapping  needs  context & approximation

24

Results with multiple ontologies

Monotonic improvement Independent of order Linear increase of cost

Separate

Lexical ICD-10 DICE MeSH

RecallPrecision

64%95%

64%95%

76%94%

88%89%

Joint

0102030405060708090

100

Lexical ICD-10 DICE MeSH

Page 24: Ontology mapping  needs  context & approximation

does structured context knowledge help?

Page 25: Ontology mapping  needs  context & approximation

26

Exploiting structure CRISP: 700 concepts, broader-than MeSH: 1475 concepts, broader-than FMA: 75.000 concepts, 160 relation-types

(we used: is-a & part-of)

CRISP (738)

MeSH (1475)

FMA (75.000)

anchoring anchoring

mapping

inference

Page 26: Ontology mapping  needs  context & approximation

27

(S <a B) & (B < B’) & (B’ <a T) ! (S <i T)

a a

i

Using the structure or not ?

Page 27: Ontology mapping  needs  context & approximation

28

(S <a B) & (B < B’) & (B’ <a T) ! (S <i T)

No use of structure Only stated is-a & part-of Transitive chains of is-a, and

transitive chains of part-of Transitive chains of is-a and part-of One chain of part-of before

one chain of is-a

Using the structure or not ?

Page 28: Ontology mapping  needs  context & approximation

29

Examples

Page 29: Ontology mapping  needs  context & approximation

30

Examples

Page 30: Ontology mapping  needs  context & approximation

31

Matching results (CRISP to MeSH)

Recall = · ¸ total incr.Exp.1:DirectExp.2:Indir. is-a + part-ofExp.3:Indir. separate closuresExp.4:Indir. mixed closuresExp.5:Indir. part-of before is-a

448

395

395

395

395

417516933151

1972

156405140

2222

8180

0

1021

1316

2730

4143

3167

-29%

167%306%210%

Precision = · ¸ total

correct

Exp.1:DirectExp.4:Indir. mixed closuresExp.5:Indir. part-of before is-a

171414

183937

35950

38112101

100%94%

100%

(Golden Standard n=30)

Page 31: Ontology mapping  needs  context & approximation

32

Three obvious intuitions

Ontology mapping needs background knowledge

Ontology mapping needs approximation

The Semantic Web needs ontology mapping

young researcher

?≈ post-doc

Page 32: Ontology mapping  needs  context & approximation

This work withZharko Aleksovski

Risto Gligorov Warner ten Kate

Page 33: Ontology mapping  needs  context & approximation

34

B

Approximating subsumptions(and hence mappings)

query: A v B ?B = B1 u B2 u B3 A v B1, A v B2, A v

B3 ?

B1 B3

B2

A

Page 34: Ontology mapping  needs  context & approximation

35

Approximating subsumptions

Use “Google distance” to decide which subproblems are reasonable to focus on

Google distance

wheref(x) is the number of Google hits for xf(x,y) is the number of Google hits for the tuple of search items x and yM is the number of web pages indexed by Google

)}(log),(min{loglog),(log)}(log),(max{log),(

yfxfMyxfyfxfyxNGD

≈ symmetric conditional probability

of co-occurrence

≈ estimate of semantic distance

≈ estimate of “contribution” to B1 u B2 u B3

Page 35: Ontology mapping  needs  context & approximation

37

Google distance

animal

sheep cow

madcow

vegeterian

plant

Page 36: Ontology mapping  needs  context & approximation

38

Google for sloppy matching Algorithm for A v B (B=B1 u B2 u

B3)

determine NGD(B, Bi)=i, i=1,2,3 incrementally:

• increase sloppyness threshold • allow to ignore A v Bi with i ·

match if remaining A v Bj hold

Page 37: Ontology mapping  needs  context & approximation

39

Properties of sloppy matchingWhen sloppyness threshold goes up,

set of matches grows monotonically=0: classical matching=1: trivial matching

Ideally: compute i such that: desirable matches

become true at low undesirable matches

become true only at high Use random selection of Bi as baseline

?

Page 38: Ontology mapping  needs  context & approximation

40

CDNow (Amazon.com)

All Music Guide

MusicMoz

ArtistGigs

Artist Direct NetworkCD baby

YahooSize: 96 classesDepth: 2 levels

Size: 2410 classesDepth: 5 levels Size: 382 classes

Depth: 4 levels

Size: 222 classesDepth: 2 levels

Size: 1073 classesDepth: 7 levels

Size: 465 classesDepth: 2 levels

Size: 403 classesDepth: 3 levels

Experiments in music domain

very sloppy terms good

Page 39: Ontology mapping  needs  context & approximation

7 16-05-2006

Experimentpr

ecis

ion

recall

classicalrandomNGD

Manual Gold Standard, N=50, random pairs

20

=0.5397

60 =0.5

Page 40: Ontology mapping  needs  context & approximation

wrapping up

Page 41: Ontology mapping  needs  context & approximation

43

Three obvious intuitions

Ontology mapping needs background knowledge

Ontology mapping needs approximation

The Semantic Web needs ontology mapping

Page 42: Ontology mapping  needs  context & approximation

44

So that shared context & approximation

make ontology-mapping a bit more like a social conversation

Page 43: Ontology mapping  needs  context & approximation

45

Future: Distributed/P2P setting

source target

background knowledge

anchoring anchoring

mapping

inference

Page 44: Ontology mapping  needs  context & approximation

46

Vragen & discussie

[email protected]://www.cs.vu.nl/~frankh