Rich Data? Poor Data? Depends on...

15
| 15 | Rich Data? Poor Data? | September 4, 2013 1 Rich Data? Poor Data? Depends on… Lars G. Svensson

description

A presentation at the workshop "Rich and loonely or poor and popular?" at the Dublin Core conference in Lisbon on September 4th, 2013. The main hypothesis is that when publishing (linked) data, the main criteria should not be richness and poorness, but suitability for purpose, granularity and adherence to agreed-on models.

Transcript of Rich Data? Poor Data? Depends on...

Page 1: Rich Data? Poor Data? Depends on...

| 15 | Rich Data? Poor Data? | September 4, 20131

Rich Data? Poor Data? Depends on…Lars G. Svensson

Page 2: Rich Data? Poor Data? Depends on...

Libraries and other cultural heritage institutions have high-quality data

| 15 | Rich Data? Poor Data? | September 4, 20132

Page 3: Rich Data? Poor Data? Depends on...

Metadata specialists are convinced that this data could make (web) search better

| 15 | Rich Data? Poor Data? | September 4, 20133

There it is!

WOW!

Ph

oto

by W

en

dt

Com

mon

s (C

C B

Y):

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/w

en

dt-

libra

ry/5

19

09

02

31

6/

Page 4: Rich Data? Poor Data? Depends on...

Linked data is about connecting silos (but those silos are not the problem…)

| 15 | Rich Data? Poor Data? | September 4, 20134

Foto

von

Doc

Searl

s (C

C B

Y):

h

ttp

://w

ww

.flic

kr.c

om

/ph

oto

s/d

ocs

earl

s/5

50

07

14

14

0/

DNB OCLCBLBnF

Page 5: Rich Data? Poor Data? Depends on...

| 15 | Rich Data? Poor Data? | September 4, 20135

… they are just a proxy for another set of silos (the model ones)

Foto

von

Doc

Searl

s (C

C B

Y):

h

ttp

://w

ww

.flic

kr.c

om

/ph

oto

s/d

ocs

earl

s/5

50

07

14

14

0/

MARC 21 FRBR

Bib-Frame

UNIMARC

Page 6: Rich Data? Poor Data? Depends on...

And that puts the burden on the consumer (making them less prone to use our data)

| 15 | Rich Data? Poor Data? | September 4, 20136

Lin

kin

g O

pen

Data

clo

ud

dia

gra

m,

by R

ich

ard

Cyg

an

iak

an

d A

nja

Jen

tzsc

h.

htt

p:/

/lod

-clo

ud

.net/

Page 7: Rich Data? Poor Data? Depends on...

We need to talk about models, not only about formats

| 15 | Rich Data? Poor Data? | September 4, 20137

Ph

oto

by W

on

derl

an

e (

CC

BY):

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/w

on

derl

an

e/4

63

28

79

97

6/

Page 8: Rich Data? Poor Data? Depends on...

The model underlying the GND (and thus the GNDO) is entity-based

| 15 | Rich Data? Poor Data? | September 4, 20138

• Corporate Body• Conference or Event• Subject Heading• Work• Place or Geographic Name• Family• Person

Page 9: Rich Data? Poor Data? Depends on...

For bibliographic resources we currently have no model (but a common German AP!)

| 15 | Rich Data? Poor Data? | September 4, 20139

<rdf:Description rdf:about="http://d-nb.info/870009036"><rdf:type rdf:resource="http://purl.org/ontology/bibo/Document" /><dcterms:medium rdf:resource="http://rdvocab.info/termList/RDACarrierType/1044" /><owl:sameAs rdf:resource="http://hub.culturegraph.org/resource/DNB-870009036" /><bibo:isbn10>3887456335</bibo:isbn10><dc:identifier>(OColc)74784457</dc:identifier><dc:title>Das Handbuch für IBM PC und Kompatible</dc:title><dcterms:creator rdf:resource="http://d-nb.info/gnd/111166799" /><rda:placeOfPublication>Düsseldorf</rda:placeOfPublication><rda:placeOfPublication>Berkeley</rda:placeOfPublication><rda:placeOfPublication>Paris</rda:placeOfPublication><dc:publisher>Sybex</dc:publisher><rda:publicationStatement>Paris : Sybex, 1986</rda:publicationStatement><isbd:P1053>332 S.</isbd:P1053><dcterms:subject rdf:resource="http://d-nb.info/gnd/4026436-1" /><dcterms:issued>1986</dcterms:issued>

</rdf:Description>

Page 10: Rich Data? Poor Data? Depends on...

We can have several models for different use cases or communities (and expressed in different formats/syntaxes)

| 15 | Rich Data? Poor Data? | September 4, 201310

Thing

(It doesn‘t have to be like this…)

Somehow related to

*

*

Page 11: Rich Data? Poor Data? Depends on...

OWL/RDF reasoning can take care of semantic granularity only if the models are compatible enough

| 15 | Rich Data? Poor Data? | September 4, 201311

htt

p:/

/doss

ierd

oc.

typ

ep

ad

.com

/.a/6

a0

0d

83

45

29

93

c69

e2

01

28

76

98

b9

ff9

70

c-p

op

up

Page 12: Rich Data? Poor Data? Depends on...

So is the solution to build another great model?

| 15 | Rich Data? Poor Data? | September 4, 201312Ph

oto

by T

han

gara

j Ku

mara

vel (C

C B

Y):

h

ttp

://w

ww

.flic

kr.c

om

/ph

oto

s/ku

mara

vel/8

15

29

74

24

6/

Page 13: Rich Data? Poor Data? Depends on...

We don‘t need one great model, but several small ones (those we can turn)

| 15 | Rich Data? Poor Data? | September 4, 201313

Ph

oto

by Joh

n D

rin

kwate

r (C

C B

Y N

C-S

A):

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/jo

hn

-r-d

/65

37

39

94

43

/

Page 14: Rich Data? Poor Data? Depends on...

After all connecting is about the connectors (and we can do it better)

| 15 | Rich Data? Poor Data? | September 4, 201314

Ph

oto

s b

y y

uan

kuei (C

C B

Y-N

C-N

D):

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/p

lease

/18

42

75

30

9/;

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/p

lease

/18

42

76

06

4/;

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/p

lease

/18

42

77

15

6/;

htt

p:/

/ww

w.fl

ickr

.com

/ph

oto

s/p

lease

/18

42

77

06

3/

Page 15: Rich Data? Poor Data? Depends on...

Don‘t let the cloud obstruct the sun!

| 15 | Rich Data? Poor Data? | September 4, 201315

Lin

kin

g O

pen

Data

clo

ud

dia

gra

m,

by R

ich

ard

Cyg

an

iak

an

d A

nja

Jen

tzsc

h.

htt

p:/

/lod

-clo

ud

.net/