+ All Categories
Home > Documents > Consensus building workshop

Consensus building workshop

Date post: 12-Jan-2016
Category:
Upload: alyn
View: 69 times
Download: 0 times
Share this document with a friend
Description:
Consensus building workshop. Conference track. Outline. Introduction (ideas behind the track) Evaluation Discussion – interesting mappings. Conference track - Features. Broadly understandable domain Conference Organisation Free exploration by participants within 10 ontologies - PowerPoint PPT Presentation
34
1 Consensus building workshop Conference track
Transcript
Page 1: Consensus building  workshop

1

Consensus building workshop

Conference track

Page 2: Consensus building  workshop

2

Outline

• Introduction (ideas behind the track)

• Evaluation

• Discussion – interesting mappings

Page 3: Consensus building  workshop

3

Conference track - Features

• Broadly understandable domain Conference Organisation

• Free exploration by participants within 10 ontologies

• No a priori reference alignment

• Participants: 6 research groups

Page 4: Consensus building  workshop

4

Conference track - Dataset

OWL, tool Protege

Page 5: Consensus building  workshop

5

Conference track - Participants

• 6 participants– Automs– Coma++– OWL-CtxMatch– Falcon– HMatch– RiMOM

Page 6: Consensus building  workshop

6

Conference track - Goals

• Focus on interesting mappings and unclear mappings– Why should they be mapped?

• Arguments: against and for

– Which systems did discover them?– Differences in similarity measures

• Underlying techniques?

Page 7: Consensus building  workshop

7

Outline…

• Introduction (ideas behind the track)

• Evaluation

• Discussion – interesting mappings

Page 8: Consensus building  workshop

8

Evaluation

• Processing all mappings by hand

• Assessment based on personal judgement of organisers (consistency problem)

• Tags: TP, FP, interesting, ?, heterogenous mapping

• Types of errors and phenomena: – subsumption, inverse property, siblings,

lexical confusion

Page 9: Consensus building  workshop

9

Evaluation…

• Subsumption mistaken for equivalence– Author,Paper_Author– Conference_Trip, Conference_part

• Inverse property– has_author,authorOf

• Siblings mistaken for equivalence– ProgramCommittee,Technical_commitee

• Lexical confusion error– program,Program_chair

• Relation – Class mapping– has_abstract,Abstract– Topic,coversTopic; read_paper,Paper

Page 10: Consensus building  workshop

10

Evaluation…

• Some statistics as a side-effect of processing

Page 11: Consensus building  workshop

11

Evaluation…

Page 12: Consensus building  workshop

12

Evaluation…

• Distribution of similarity measures – for True Positive Mappings and – for False Positive Mappings

Page 13: Consensus building  workshop

13

Evaluation…Coma++

0

50

100

150

200

250

300

350

400

450

500

0-20% 20-40% 40-60% 60-80% 80-100%

TP

FP

Coma++

0

50

100

150

200

250

300

350

400

450

500

0-20% 20-40% 40-60% 60-80% 80-100%

TP

FP

Page 14: Consensus building  workshop

14

Evaluation…

Falcon

0

50

100

150

200

250

300

350

400

0-20% 20-40% 40-60% 60-80% 80-100%

TP

FP

Page 15: Consensus building  workshop

15

Evaluation…

HMatch

0

50

100

150

200

250

300

350

400

0-20% 20-40% 40-60% 60-80% 80-100%

TP

FP

Page 16: Consensus building  workshop

16

Evaluation…RiMOM

0

100

200

300

400

500

600

700

0-20% 20-40% 40-60% 60-80% 80-100%

TP

FP

RiMOM

0

100

200

300

400

500

600

700

0-20% 20-40% 40-60% 60-80% 80-100%

TP

FP

Page 17: Consensus building  workshop

17

Record it!

Page 18: Consensus building  workshop

18

Outline…

• Introduction (ideas behind the track)

• Evaluation

• Discussion – interesting mappings

Page 19: Consensus building  workshop

19

Discussion

• Focus on interesting mappings and unclear mappings– Why should they be mapped?

• Arguments: against and for

– Which systems discover them?– Differences in similarity measures

• Underlying techniques?

Page 20: Consensus building  workshop

20

Mapping 1

Element1 Element2 Relation

Person Confious:human =

Notes semantically same

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

1.0 Iasted

1.0 Ekaw

No No No 0.7 confOf

0.63 Ekaw

0.81 Sigkdd

0.77 sofsem

1.0 PCS

Page 21: Consensus building  workshop

21

Mapping 1confiousIasted

confOf

ekaw

sigkdd

sofsem

PCS

Page 22: Consensus building  workshop

22

Mapping 2

Element1 Element2 Relation

OpenConf:Surname confious:last_name =

Notes Both are datatype properties, the former with People as domain, the latter with human as domain

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

1.0 No No No 1.0 No

Page 23: Consensus building  workshop

23

Mapping 3

Element1 Element2 Relation

sofsem:has_the_last_name confious:last_name =

Notes Both are datatype properties, the former with Person as domain, the latter with human as domain

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

No 0.63 No No 0.8 No

Page 24: Consensus building  workshop

24

Mapping 4

Element1 Element2 Relation

ekaw: PC_Member confOf:Member_PC =

Notes Change order of incompound names

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

1.0 No No No No 0.53

Page 25: Consensus building  workshop

25

Mapping 4

confOf

ekaw

Page 26: Consensus building  workshop

26

Mapping 5

Element1 Element2 Relation

ekaw:Document confious:article =

Notes Semantically same?

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

1.0 No No No No 1.0

Page 27: Consensus building  workshop

27

Mapping 5

confious

ekaw

Page 28: Consensus building  workshop

28

Mapping 8

Element1 Element2 Relation

cmt:Rejection OpenConf:Reject =

Notes Both relates to process of assessment.

But the former is a recommendation, the latter is a decision. So…

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

No 0.29 No 0.94 No No

Page 29: Consensus building  workshop

29

Mapping 8

cmt

OpenConf

Page 30: Consensus building  workshop

30

Mapping 11

Element1 Element2 Relation

ekaw:Location Place =

Notes Semantically same? Both are at the highest level of hierarchy. But Location maybe more general than Place… what about City?

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

No No No No iasted 0.8

sigkdd 0.8No

Page 31: Consensus building  workshop

31

Mapping 11ekaw

iasted

Asserted conditions for iasted:Place

Location is domain of properties: locationOfLocation is range of properties: heldIn

iasted:Place is domain of properties: is_equipped_bysigkdd:Place is range of properties: can_stay_in

sigkdd

Page 32: Consensus building  workshop

32

Mapping 12

Element1 Element2 Relation

sofsem:reviews ekaw:hasReview =

Notes DomainOf(hasReview)=Paper,rangeOf(hasReview)=Review

DomainOf(reviews)=Review,rangeOf(reviews)=Reviewed_contribution

Inverse property phenomena, useful?

System

Automs Coma++ OWL-CtxMatch

Falcon HMatch RiMOM

No No 1.0 No No No

Page 33: Consensus building  workshop

33

• Call for contribution to our dataset

Page 34: Consensus building  workshop

34

Thank you for your participation!


Recommended