Database Reference
In-Depth Information
INTRODUCTION
these 25 countries is returned also for the years
before their membership. But of course, the
overall sum of returned inhabitants for the year
1987 does not match the real number of people
living in the European Union at that time. Com-
paring the numbers of 1990 and 1991, where the
organization itself did not change, may indicate
a massive increase of inhabitants. In reality, the
1991 number also contains the 16.4 million people
of former East-Germany. Eurostat, for instance,
takes this into account, and presents numbers
of the united Germany also for the years before
1991. Another example for an unclear inclusion
are the Baltic countries or Slovenia. They did not
even exist before 1991, but were parts of other
countries, which, of course, never were parts of
the European Union. An alternative to presenting
such “adjusted data” is to display the “historical
truth”, i.e. include the numbers of different coun-
tries only after they joined the European Union.
This may make sense in some situations, in others,
such results may be useless.
An example demonstrating the effect of
changing semantics could be to retrieve the Gross
National Product of the countries in the European
Union from 1983 to 2008. Besides the problems
induced by the structural changes described
above, i.e. whether and how to include numbers
for a specific country, this query illustrates the
changing semantics problem: As of 1999 and
2002, a common European currency, the Euro,
was introduced as deposit currency and cash
money respectively, in many - but not all - of
the member countries. Thus, before 1999 the
Gross National Product of different countries was
expressed in the local currency, but as of 1999 it
is given in Euro. Before 1999 for comparing the
GNP of different countries, it is obvious that the
numbers must be brought to a common base, i.e.
the same currency, to be comparable. But what
about statistics for a single country? ForAustria, 1
Euro exchanges 13.7603 Austrian Schillings. So,
someone comparing the Austrian Gross National
Product from 1990-2006 without considering the
The standard architecture for data warehouse
systems are
multidimensional databases
, where
transactional data
(
cell values
) are described in
terms of master data (
dimensions
and
dimension
members
). Whereas today's commercial systems
are well prepared to deal with changes in the
transactional data, they are, surprisingly, not able
to deal with changing master data in a satisfactory
way. Nonetheless, such changes frequently occur
(restructuring in organizations, new laws, mergers
and acquisitions, product portfolio restructuring,
etc.). All these changes have to be represented in
the information systems, and thus, must some-
how be modeled also in the data warehouse. For
data warehouses the adequate representation and
treatment of such changes is even more crucial
than in standard database applications, since data
warehouses are intended to represent also histori-
cal data which - changes occurring - might be
quite incompatible.
A simple example illustrating the problem of
missing data is querying the number of inhabit-
ants in the European Union for the last 25 years.
This query seems rather straightforward and the
numbers should not leave much space for interpre-
tation. But, one has to be aware of some changes:
First of all, the geopolitical entity “European
Union” only exists since 1993, succeeding the
“European Community”, which itself was origi-
nally named “European Economic Community”.
Furthermore, in the considered period (1983 to
2008), the European Union grew from 12 to 27
members. Finally, with the reunification of East-
and West-Germany in 1990 one of the member
countries had a massive internal reorganization.
So if querying the number of inhabitants from
1983 to 2008, how can the resulting numbers
be compared? When querying this data from the
Eurostat website, one has to choose the “geopoliti-
cal entity” (EU-27, EU-25, one or more counties,
…) for which the data should be retrieved. If,
for instance, EU-25 is chosen, the population for
Search WWH ::
Custom Search