Statistical databases: the reference environment and three layers proposed by Eurostat
Abstract
The functions of the Eurostat information system are divided into four sectors which correspond to the various stages in the processing of data from their collection to their difussion: - Production: collection, validation and storage of the data and meta-data; - Storage of the reference data (acceptance of the information); - Use of the reference data (visibility/security and find/deliver); - Diffusion. The system of acquisition and validation represents for Eurostat a major workload given the diversity of the partners and of the different data processing situations encountered. The upshot of the diversity and change in both the production and dissemination environments is the need to de-couple them. If there are n production systems and m dissemination systems, then there will be a need for n*m interfaces between them. This will lead to an explosion of work to long term to have the data available to end-users. If a reference layer is inserted to de-couple the production and dissemination environments, the interface problem reduces to n+m and therefore becomes much more manageable. Once the interfaces to and from the reference layer are defined, new tools, products or domains in either environment are easily added. By de-coupling production from dissemination, a reference layer enables producers the freedom that they require to choose the most appropriate tools an techniques to do their job. Producers also gain a common interface into the reference area, rather than the many different interfaces that they have to cope with at present. By providing well documented data in a standard form, the reference layer provides disseminators with better raw materials for their products. This enables them to provide better and more diverse products in a more timely and customer-driven fashion, as desired in Eurostat's mission statement.
Full text
Q¨
UESTII ´
O, ol. 21, 1 i 2, p. 233-240, 1997
STATISTICAL DATABASES: THE REFERENCE
ENVIRONMENT AND THREE LAYERS PROPOSED
BY EUROSTAT
ROGER DUBOIS
Eu os a
The unc ions o he Eu os a in o ma ion sys em a e di ided in o ou sec o s which
co espond o he a ious s ages in he p ocessing o da a om hei collec ion o
hei di usion:
– P oduc ion: collec ion, alida ion and s o age o he da a and me a-da a;
– S o age o he e e ence da a (accep ance o he in o ma ion);
– Use o he e e ence da a ( isibili y/secu i y and ind/deli e );
– Di usion.
The sys em o acquisi ion and alida ion ep esen s o Eu os a a majo wo klo-
ad gi en he di e si y o he pa ne s and o he di e en da a p ocessing si ua ions
encoun e ed. The upsho o he di e si y and change in bo h he p oduc ion and dis-
semina ion en i onmen s is he need o de-couple hem. I he e a e n p oduc ion
sys ems and m dissemina ion sys ems, hen he e will be a need o n
m in e aces
be ween hem. This will lead an explosion o wo k o long e m o ha e he da a
a ailable o end-use s. I a e e ence laye is inse ed o de-couple he p oduc ion
and dissemina ion en i onmen s, he in e ace p oblem educes o n
+
m and he-
e o e becomes much manageable. Once he in e aces o and om he e e ence
laye a e de ined, new ools, p oduc s o domains in ei he en i onmen a e easily
added. By de-coupling p oduc ion om dissemina ion, a e e ence laye enables
p oduce s he eedom ha hey equi e o choose he mos app op ia e ools and
echniques o do hei job. P oduce s also gain a common in e ace in o he e e-
ence a ea, a he han he many di e en in e aces ha hey ha e o cope wi h a
p esen . By p o iding well documen ed da a in a s anda d o m, he e e ence la-
ye p o ides dissemina o s wi h be e aw ma e ial o hei p oduc s This enables
hem o p o ide be e and mo e di e se p oduc s in a mo e imely and cus ome -
d i en, as desi ed in Eu os a s mission s a emen .
Keywo ds: A chi ec u e, e e ence en i onmen , s a is ical in o ma ion
sys em.
*Roge Dubois. Adminis a o Uni A3. In o ma ion Da a-Shop. Eu os a . L-2920 Luxembou g.
–A icle ebu l’oc ub e de 1996.
–Accep a el maig de 1997.
233
1. FUNCTIONAL OBJECTIVES OF THE INFORMATION SYSTEM
The unc ions o he Eu os a in o ma ion sys em a e di ided in o ou sec o s
which co espond o he a ious s ages in he p ocessing o da a om hei collec ion
o hei di usion:
– p oduc ion: collec ion, alida ion and s o age o he da a and me a-da a;
– s o age o he e e ence da a (accep ance o he in o ma ion);
– use o he e e ence da a ( isibili y/secu i y and ind/deli e );
– di usion.
Co-ope a ion be ween hese ou specialised sec o s in ol es gene alised unc ions o
ca aloguing and con ol (manipula ion o da a/ esou ces managemen ).
In e ms o hei na u e, he unc ions can be classi ied in o six ca ego ies:
– acquisi ion and alida ion o he da a in o six ca ego ies;
– managemen o he p oduc ion da a;
– manipula ion o he da a and me a-da a;
– ex e nal p esen a ion;
– associa ed se ices.
The sys em o acquisi ion and alida ion ep esen s o Eu os a a majo wo kload
gi en he di e si y o he pa ne s and o he di e en da a p ocessing si ua ions
encoun e ed. The objec i e o he de elopmen o his sys em is o minimise he
wo kload equi ed despi e he inc ease in demand o s a is ical da a. The use o
EDI echniques (exchange o compu e ised da a) is he di ec ion o be ollowed in an
a emp o elimina e he a ie ies o magne ic media, o o ma and o in e p e a ion
in he collec ion o da a.
The managemen o he p oduc ion da a has o allow he mig a ion o he a ious
ypes o objec s which combine o o m Eu os a s in o ma ion sys ems. The Nomen-
cla u es a e called o play an essen ial ole in he c ea ion, he homogenisa ion and
hen he documen a ion o he s a is ical da a; hey cons i u e de ac o he hea o he
in o ma ion sys em.
The manipula ion o he da a mus , in he i s place, suppo he p ocess o da a
ans e be ween he a ious en i onmen s. A numbe o se ices appea s which
allow he es ablishmen o g oups o use s and o hei associa ed p i ileges; ce ain
me a-da a become con ol ules in he in o ma ion sys em.
234
The di usion mus consis o quali y p oduc s which a e well-known and a e based
on a pla o m which is easy o acqui e and use (da abases, s a is ical documen s and
publica ions, ex ac ions om da abases, magne ic media, CD-ROM, e c.).
The las wo sys ems a e ac i i ies o a gene al se ice na u e which a e necessa y
o any da a-p ocessing a chi ec u e, while emphasising pa icula ly wo e y impo -
an ac i i ies o Eu os a : he STRINGS p ojec o elec onic publishing and he
secu i y o EDP applica ions.
2. TYPOLOGY OF THE DATA
S a is ical da a can be di ided in o wo in e dependen subjec s:
S a is ical obse a ions which co espond o all quan i a i e (and quali a i e)
measu es associa ed wi h economic phenomena. This g oup is p ima ily com-
posed o nume ical alues associa ed wi h in o ma ion on hei empo al e o-
lu ion.
The p incipal ypes o p ojec s which make i possible o o ganise and o s uc-
u e his se o da a a e:
– ec o s and ma ices o numbe s;
– ime se ies;
– ables wi h one o mo e dimensions.
All he s a is ical me a-da a ela ing o he cha ac e isa ion o economic pheno-
mena and o s a is ical obse a ions.
An i em o s a is ical me a-da a is cha ac e is ically associa ed wi h one o a
collec ion o s a is ical obse a ions so as o speci y he signi icance o he
ela ed da a.
The p incipal ypes o objec s which cha ac e ise me a-da a a e:
– The Nomencla u es and he dimensions, which allow he o ganisa ion, he clas-
si ica ion and he associa ion o economic phenomena and s a is ical obse a-
ions.
– The p ope ies and documen a ion, which allow he cha ac e isa ion and he
documen a ion o he phenomena and he obse a ions.
235
Models: which de ine he p ocesses (o de i a ion o o alida ion) on he
s a is ical obse a ions.
I is e y di icul o gi e a single de ini ion o he me a-da a concep and i s ep e-
sen a ion o he ollowing easons:
– The me a-da a se es as a link be ween a conc e e en i y, he s a is ical obse -
a ion, and an abs ac en i y he economic phenomenon associa ed wi h he
obse a ion.
No e: he empo al e olu ion o hese me a-da a is he complexi y ac o .
– The me a-da a concep may be used ei he « o classi y»o « o seek»a subse s
o obse a ions. The hypo heses which unde lie hese wo ac i i ies can be
di e en and gi e ise o speci ic me a-da a.
– In addi ion o he «link»aspec , me a-da a cha ac e ise he «p ope ies»o a
subse o in o ma ion.
Fo example, he me a-da a associa ed wi h a ime se ies a e:
=
)
he economic phenomenon whose e olu ion is de ined by he se ies;
=
)
he equency, he sou ce, he ype o se ies ( low, s ock, e c.), he uni ,
he o de o magni ude, e c.;
=
)
one o mo e commen s o desc ip i e ex s on one o o he aspec o he
se ies.
3. ARCHITECTURAL SOLUTION: A REFERENCE LAYER
3.1. A chi ec u e
The upsho o he di e si y and change in bo h he p oduc ion and dissemina ion
en i onmen s is he need o de-couple hem. I he e a e np oduc ion sys ems and
mdissemina ion sys ems, hen he e will be a need o n
min e aces be ween hem.
This will lead o an explosion o wo k o long ime o become a ailable o end-use s.
This is no in line wi h Eu os a s s a ed objec i e o imely p o ision o da a.
I a e e ence laye is inse ed o de-couple he p oduc ion and dissemina ion
en i onmen s, he in e ace p oblem educes o n
+
mand he e o e becomes much
mo e manageable. Once he in e aces o and om he e e ence laye a e de ined,
new ools, p oduc s o domains in ei he en i onmen a e easily added. This enables
236
Eu os a o manage he e olu ion and change inhe en in he Eu opean s a is ical
en i onmen , because changes o one pa o ei he p oduc ion o dissemina ion will
no ha e de imen al impac s o any o he pa .
MANAGEMENT
PRODUCTION
sys emsnsys emsm
DISSEMINATION
REFERENCE
➜
➜
➜
➜
➜
➜
3.2. Po en ial bene i s
A. Fo p oduce s
By de-coupling p oduc ion om dissemina ion, a e e ence laye enables p oduce s
he eedom ha hey equi e o choose he mos app op ia e ools and echniques o
do hei job. I p o ides well documen ed da a o p oduce s when hey a e using
o he domains. I also helps hem o unde s and hei own domains when hey look
a old da a in hei own domain, no only because hei own memo ies may no be
comple ely accu a e, bu because he mo emen o p oduce s wi hin he o ice means
ha expe ise is no always a ailable.
P oduce s also gain a common in e ace in o he e e ence a ea, a he han he
many di e en in e aces ha hey ha e o cope wi h a p esen . They ha e mo e
mobili y o o he jobs wi h Eu os a , because he common wo king en i onmen will
be amilia o hem in a new pos .
B. Fo dissemina o s
By p o iding well documen ed da a in a s anda d o m, he e e encelaye p o ides
dissemina o s wi h be e aw ma e ial o hei p oduc s. This enables hem o p o ide
be e and mo e di e se p oduc s in a mo e imely and cus ome -d i en manne , as
desi ed in Eu os a s mission s a emen .
237
C. Fo cus ome s
While no di ec ly in ol ed in he e e ence laye , he e e ence laye should
p o ide cus ome s wi h da a ha is:
– Visible. They will know wha is a ailable.
– Unde s andable. I will be documen ed o an ag eed minimum s anda d.
– Accessible. They will be able o ge he da a mo e easily.
– Compa able. he s anda disa ion o me a-da a will help o enable c oss-domain
compa ison.
– Timely. The da a should be a ailable mo e quickly.
– Accu a e. The au ho isa ion o da a mo ing ou o p oduc ion should ensu e i s
high quali y.
– App op ia e. Feedback on usage by use s should help Eu os a p o ide wha
use s need.
D. Fo manage s
A e e ence laye assis s manage s because i can p o ide:
– Cen al secu i y and access con ol o Eu os a da a. This ensu es ha da a is
no le on insecu e and ulne able PC pla o ms, and can be sa egua ded as
he i al esou ce ha i is.
– Ha monisa ion o nomencla u es and code lis s leading o less duplica ion o
e o in cap u e and ansla ion o me a-da a. P oduce s will be able o e-use
p e ious wo k by o he s. The e will also be inc eased compa abili y o da a be-
ween di e en domains because o ha monised de ini ions o hei dimensions.
– Flexibili y o choice in p oduc ion ools and me hods and in dissemina ion
p oduc s.
– S abili y o Eu os a da a and me a-da a o e ime.
– Co-o dina ed ma ke eedback o enable Eu os a o dissemina e wha hei cus-
ome s wan and need. This in o ma ion will come om bo h he e e ence and
dissemina ion en i onmen s.
238
4. TYPOLOGY OF THE PROCESSES
The p ocesses co esponding o one o o a numbe o unc ionali ies, mus occu
in one o he de ined en i onmen s: p oduc ion, e e ence, di usion.
An EDP applica ion is seen as an o de ed se o p ocesses; he la e need no
necessa ily be on he same compu e , bu can be on wo (o se e al) machines o he
same o di e en pla o ms.
The dis ibu ion o he p ocessing wi hin an applica ion o e he di e en pla o ms
is a di icul decision o which some ules ha e o be ixed:
=
)
gene al ules applicable h oughou he Eu opean Union;
=
)
olume conside a ions conce ning he da a and he popula ion o po en ial
use s;
=
)
secu i y conside a ions equi ing a mo e o less p o ec ed en i onmen ,
access con ol unc ions, backup and eco e y, a chi ing e c.;
=
)
he a ailabili y o sui able so wa e on he di e en pla o ms.
4.1. P oduc ion
The p oduc ion laye di ides in o se e al subjec s acco ding o he ype o objec s
managed; p ecise desc ip ion o he objec s is needed and depends on he cons uc-
ion o s anda d applica ions gi ing a amewo k o u ilisa ion o Eu os a s a ious
p oduce s. Eu os a has selec ed, on his basis, FAME o se ies and ACUMEN o
mul idimensional objec s.
4.2. Re e ence
The «accep ance o in o ma ion»poses he p oblem o dynamic de ini ions and he
au oma ed con ol o he passage be ween p oduc ion and he e e ence. I depends
on he p oduc ion s uc u es and on he complexi y o he ules o be implemen ed. In
iew o he olume o exis ing in o ma ion, only a comple ely au oma ed p ocedu e
is iable.
The concep o quali y o managed in o ma ion is i al o Eu os a : he sepa a ion
in o wo en i onmen s mus p ese e he weal h o exis ing in o ma ion, i.e. allow
mo e ho ough in e oga ion o he se o in o ma ion a ailable i ha is necessa y.
The managemen o e e ence is conce ned wi h a homogenous solu ion allowing
uni o m access o he se o managed in o ma ion.
239
The isibili y and secu i y laye mus ensu e con ol o he ac i i ies o use s
p i ileges and o da a con iden iali y. I is impo an o gua an ee ha no leaks o
in o ma ion c oss his laye o so wa e.
The «FIND and DELIVER»laye allows a non-specialis use easy na iga ion
a ound Eu os a s in o ma ion and a apid and exhaus i e eply o his ques ions.
4.3. Di usion
Eu os a dissemina es s a is ical in o ma ion by means o 3 ypes o p oduc s:
=
)
Publica ions o s a is ical epo s, which need o mix ex s, g aphs and
ables, using desk- op publishing p oduc s.
=
)
Da abases.
=
)
O he elec onic di usion se ices (download, CD-ROM, e c.).
240