Structural Metadata - Abu Dhabi Statistics Conference 2013 - SCAD

religiondressInternet και Εφαρμογές Web

21 Οκτ 2013 (πριν από 4 χρόνια και 20 μέρες)

72 εμφανίσεις

Metadata


A Bedrock for Official Statistics

Dr. S. M. Tam, Chief Methodologist, ABS

Outline


What questions am I addressing?



What metadata to use to support dissemination?


How much metadata to use?


-

The answer Is ……….. for the paper world; and …….for the WWW.



Metadata


A “boring” and “confusing” subject?


“Data about data”?



Metadata for


Fitness of purpose (Reference metadata)


information exchange


humans and WWW (Structural metadata)


Improving efficiency of processes (Process metadata)


will not cover in this talk




Reference metadata


“Data on fitness for purpose”




An example from
ABS 2006 Census


It exemplifies


“good practice” to hyper
-
link up information


Conceptual metadata


Production metadata (
eg

Quality declarations)





3,380

67.1

255,052

71.4

15,017,847

69.8

155

3.1

2,956

0.8

185,039

0.9

138

2.7

13,050

3.7

911,593

4.2

127

2.5

6,592

1.8

318,969

1.5

100

2.0

5,886

1.6

295,362

1.4

79

1.6

2,4242

0.7

171,234

0.8

What do these numbers signify?


Country of Birth

Florey

%

Australian Capital
Territory

%

Australia

%

Australia

3,380

67.1

255,052

71.4

15,017,847

69.8

Vietnam

155

3.1

2,956

0.8

185,039

0.9

England

138

2.7

13,050

3.7

911,593

4.2

China (excludes SARs and Taiwan)

127

2.5

6,592

1.8

318,969

1.5

India

100

2.0

5,886

1.6

295,362

1.4

Philippines

79

1.6

2,4242

0.7

171,234

0.8

In Florey (State Suburbs), 67.1%

of people were born in Australia. The most common countries of birth were Vietnam 3.1%, England 2.7%, China
(excludes SARs and Taiwan) 2.5%,
India 2.0% and Philippines 1.6%.


Numbers will only have meaning if they have context


Structural metadata


“Data about content and container”


provides the context


for human consumption


for machine
-
to
-
machine communication



Structural metadata can be described by


Container :Dimensions (variables)


Content: Attributes (observations)


Content: Measures (units of measurement)

Data Set Structure: Concept Usage

Unit Multiplier

Unit

Topic

Time/Frequency

Country

Stock/Flow

Observation

(Dimension)

(Dimension)

(Dimension)

(Attribute)

(Dimension)

(Dimension)

(Attribute)

(Measure)

Structural metadata used to support


Discovery of official statistics


Search engines




“Linked Data”


Semantic Web/Web 3.0



Data visualisation




Machine to machine communication



Technical standards for structural metadata


Statistical Data and Metadata Exchange (SDMX)


Data Documentation Initiative (DDI)


Data Cube Vocabulary (DCV


W3C)

Structural metadata to discover statistics


Discovery metadata


Variable names (Container Structural Metadata)



Other means (specially created) to aid
WWW search


Key words


Catalogues
etc.



Google search


“Page rank” to rank matches based on


Frequency of keywords on webpage


Age of webpage


No. of other sites linking to the webpage


SEO is a big industry


Search engines do not always provide meaningful

answers


How many Web 3.0 companies are there in Abu Dhabi?


143 million hits from Google search


Yet there are only …. companies from SCAD



A “deficiency” of Web 2.0


C
ontent of the structural metadata is NOT the problem


Need relationship between “objects” recorded on the web, and query technologies



So a new approach Is needed


Web 3.0 or web of linked data


Brain child of Sir Tim Berners
-
Lee



Structural metadata to support “linked data”


What is the
difference between
Web 2.0 and Web
3.0?


What is linked data?


Linked data


Tim Berners
-
Lee


5 star Open Data Format



In a nutshell


Structure the “Structural metadata” using Resources Description Framework (RDF)


Identity statistical concepts in Universal Resources Identifiers (URIs)


GSIM uniquely identifies metadata objects



Linked data is an emerging but an increasingly important field for official statistics


Help us “ingest” data better


Help other better “digest” our data


USBC, CSO Ireland, Statistics Switzerland etc. have trialled linked data


SemStat

2013 to be held Sydney, Australia


Structural metadata to support ……..


Data visualisation (DV)


Structural metadata harvested for the visualisation application


SDMX converter for Google Public Data Explorer


DV applications built on/support SDMX


Flex


CB


NCOMVA’s Statistics
eXplorer



Exchange of statistics from one computer to another


Web Services


Structural metadata


Technical standards to describe structural metadata such as SDMX


Web Services protocols or standards


WSDL, SOAP and XML



To summarise


What
metadata to use to support dissemination?


How much metadata to use
?



Dissemination goals


Assist users to determine fitness for purpose


“Consume” the data



increasingly through linked data


Data visualisation


Machine to machine communication


-
The
answer Is
Reference Metadata
the paper world;
and


-
Reference Metadata
and

Structural Metadata

(+
S
uitable Technical Standards) for
the
WWW.


Useful references for linked data cubes


Still a new an emerging field for statistical data


General introduction of Data Cube Vocabulary


http://www.slideshare.net/der42/linked
-
data
-
hypercubes


General introduction to Linked Statistical Data


Statistical Linked
Dataspaces


A simple description of how linked data works


http://data.gov.uk/blog/what
-
is
-
linked
-
data


TED talks by Sir Tim Berners
-
Lee


Tim Berners
-
Lee on the next Web | Video on TED.com


http://blog.ted.com/2009/03/13/tim_berners_lee_web/



Questions?

Siu
-
Ming.Tam@abs.gov.au