information extracted from one or several databases. We argue that it is not
efficient to maintain two distinct XML files, one for the geometric descrip-
tion of the features and one for the association between features and infor-
mation. In fact, successful commercial products adopt the unique file
solution, which stores the association as attributes of the geometric features.
The adoption of this approach gives rise to a number of design problems that
concern how the GIS data model and XML model should be revised and
extended to keep track of the feature-information associations during the
construction of a GIS map.
How is thematic information extracted from the databases?
Thematic information can be retrieved from several relational databases.
We assume that all of these databases are accessible directly from the GIS
server, i.e. we do not consider distributed databases. Every database has a
unique logical name and is made up of a number of tables. Every table may
store thousands of records and every record may have tens of attributes.
Typically, only a subset of the entire volume of data is relevant for a
specific GIS service. Thus, it is worthwhile to find an approach to reduce the
dimension and number of tables and records that need to be considered. We
have to take into consideration the following issues:
Only a subset of the number of GIS databases needs to be considered.
For each GIS database, only a subset of its tables stores relevant infor-
mation for that service.
For each table, only a subset of its records and attributes are relevant for
We should adopt the less intrusive approach that does not require restruc-
turing the databases.
The simplest approach consists in building a new database (the GIS data-
base) that maintains only the index of the available thematic databases and
of the tables that store relevant information. Figure 15.7 depicts the table's
layout of the GIS database.
Table Databases (see Table 15.2) records the logical name ( Name ) and
description ( Description ) of a set of databases that store information useful
for the GIS service at hand. Every database is uniquely identified with the
values of the ID attribute. For each database, Tables records the list of the
tables relevant for the GIS service (see Table 15.3). Each table is described
with six attributes: a unique identifier ( ID ), a reference to the database it is
part of ( IDDataBase ), its name ( Name ), a textual description of the table
( Description ), the name of the attribute that plays the role of key ( Key ), and
the name of the attribute that plays the role of label ( Label ). Figure 15.8
shows the contents of these two tables for the example described above.