Java Reference
In-Depth Information
4.3
■
Architecture and planning
The system is very simple. It is based on low-level or algorithmic details that
will be addressed at the design level. The most relevant architectural issue is
the separation of the generic problem solution algorithms from the visual-
ization and interaction with the user. The overall architecture of the system
should be the one depicted in Figure 4.2. There are two main components:
the user interface and the classifier.
The algorithmic part of the system is confined in the
Classifier
component,
while the interaction with the user is the responsibility of the
User Interface
component.
The system will be developed in an iterative and incremental fashion.
In particular we plan to develop the system according to the following
iterations, which produce three running prototypes.
Prototype 1: Classification
. The essential classes for the classifier are
developed and the classification algorithm is implemented.
■
Prototype 2: Criteria representation
. The representation of the decision
tree is addressed.
■
Prototype 3: Training
. The tree construction algorithm is implemented,
so that a tree can be built based on a training set of items.
■
4.4
■
Prototype 1: Classification
This prototype is able to assign a category to an item based on the infor-
mation contained in a decision tree. We assume that someone has computed
the decision tree such as the one shown in Figure 4.1.
4.4.1
Analysis
We will exploit the classification algorithms defined in the literature to
develop a generic solution that can be applied both to the problem under
study (cars) and to possible extensions to other types of items.
We face two main features: first the representation of the domain, i.e. of
the items, their characteristics, and the categories they belong to; second
the definition of the classification rules.
User
interface
Classifier
Figure 4.2
Component architecture
