What Is Data Mining and How Does It Work? - Discrimination and Privacy in the Information Society - page 29

relational database . In a relational database, data is organized in tables that are
linked together. See, for instance, the following example database consisting of
three tables, Student, Course, and Grade.

Student

SID Fname Sname Dob

0001 John Williams 10/05/1985

0002 Peter Peterson 08/09/1984

0003 Ann Van Hee 07/05/1986

Course

Code Name Lecturer

2II15 Datamining T. Calders

2ID45 Databases G. Fletcher

Grade

SID Code Grade

0001 2II15 7

0002 2II15 6

Fig. 2.1 Example of a relational database consisting of three relations

Every row in a table will be called a tuple or a record , and every column
corresponds to a specific characteristic, or attribute of the tuple.
Example 5 (Relational Database). In the table Student in Figure 2.1, every tuple
corresponds to one particular student. For every student, the name (attributes
fname and sname), the student identity (attribute SID), and date of birth (attribute
dob) are recorded.
Every table has one or more attributes, that together uniquely define the identity of
the objects stored in the table. Such a specially designated combination of
attributes is called the primary key of the table. For example, in students SID is the
primary key, in courses Code is the primary key, and in the table Grade, the
combination of the attributes SID and Code form a primary key. The primary keys
are used to establish the links between the tables. For instance, in table Grade, the
primary key of Student, i.e., SID, is used to link to a particular student for which
the grade is being recorded. A good database design reduces the redundancy and
inconsistency in the data. Redundancy refers to the unnecessary repetition of
information. Suppose, for instance that next to the SID, the table grade would also
include the other attributes of students, then we would repeat a student's name and
date of birth for every course the student has a recorded grade. Not only would
this be wasteful, it would also lead to inconsistencies in the data; there could be
different tuples with the same identity, but a different date of birth. With a good
database design many of such problems can be avoided automatically.