Irrelevant Feature and Rule Removal for Structural Associative Classification Using Structure-Preserving Flat Representation - Feature Selection for Data and Pattern Recognition - page 202

Table 10.3 Integer mapping
for web pages from Table 10.2

ID Web page

0 Homepage

1 Research

2 Topics

3 51-business-intelligence

4 55-e-education-ecosystems

5 Seminars

6 413-presentation-by-eric-feinberg

7 phd-a-msc

8 Scholarships

9 scholarships.html#debii

10 About

11 Objectives

12 Mission-and-vision

13 Centres-and-labs

14 217-anti-spam-research-lab-asrl

15 214-centre-for-stringoLogsy-a-applications-csa-

16 Jobs

17 Contact-us

18 Publications

19 Conf-a-journal-papers

20 Allstaff

21 Research Professors & Fellows

22 Research-training

10.3.3 Database Structure Model (DSM)

The definition given by [ 14 ] is utilized here to describe the Database Structure Model
(DSM). Generally, the string-like representation of a tree database (example given
in Table 10.4 , is converted into a flat data format while preserving the ancestor-
descendant and sibling node relationships. Henceforth, this structure-preserving flat
data representation will be simply referred to as 'table'. The header of the table
contains the DSM without any specific attribute names. It represents only the most
general structure where every instance from the tree database can be matched to. This
will ensure that when the labels of a particular transaction from the tree database are
processed, they are placed in the correct column, corresponding to the position in
the DSM that this label matches. To illustrate the complete conversion process using
DSM, please refer to Fig. 10.2 . Using the string encoding format representation [ 51 ],
the tree database T db from Fig. 10.2 would be represented as is shown in Table 10.1 ,
where the left column corresponds to the transaction identifiers, and the right column
is the string encoding of each subtree.