Combining and Analyzing Judicial Databases - Discrimination and Privacy in the Information Society - page 189

Fig. 10.2 An overview of the data warehouse approach
object in reality. Note that this rule of thumb requires that the selectivity factors of
the common attributes are small. 7
Example: An offender-oriented data warehouse
An example of a data warehouse in the Dutch criminal law chain is the offender-
oriented data warehouse. 8 In this data warehouse, data from different judicial
databases (HKS and OM-data) were integrated by applying the rule of thumb ex-
plained above. Additionally, the data was structured and combined in such a way
that all data relate to individuals.
In the data warehouse the 'intersection' of the to-be-combined databases was
exploited. HKS stores information on three entities: suspects, the official reports
about them, and the offences of which they are suspected. OM-data also records
information about suspects and offences. Additionally, it registers case-related in-
formation. Thus, the databases were integrated based on the attributes concerning
the two common entities, that is, suspects and offences. To do so, the databases
were compared to each other and the probability that two records relate to the
same person based on common attributes was determined. While doing so, domain
knowledge was considered, for instance, the fact that an offence is usually
reported to the police on the same day as it is committed. The date of an official
report in HKS was, therefore, considered to be the same as the date of an offence
in OM-data.
As an example, assume that HKS contains a record relating to a person who
resides in Amsterdam and in respect to whom an official report has been filed
on September 1, 2010. Additionally, assume that OM-data contains a record of a
7 Choenni, S., Blanken, H. & Chang, T. (1993).
8 Choenni, S., van Dijk, J. & Leeuw, F. (2010), Choenni, S. & Meijer, R. (2011).