Data Mining Process - Java Data Mining: Strategy, Standard, and Practice - page 74

the time and money in a data warehouse. C-level 1 executives want
to get more return on their investment.
So, how does the data mining process fit into enterprise software
architectures?

3.5.1 Architectures

Data mining tools provide a wide range of options for integrating
with enterprise software architectures and processes. First, we take a
look at data mining tools themselves and then how they interact with
data sources.
As illustrated in Figure 3-10, we can view data mining tools as
consisting of four parts: a graphical user interface (GUI) for
interactive data mining, an application programming interface (API)
for building applications, a data mining engine (DME) where the
core processing or data mining algorithm execution occurs, and a
mining object repository (MOR) where persistent mining objects are
stored. Some tools may provide multiple APIs based on different
programming languages (e.g., Java, SQL, C#, C) or may involve Web
services. 2 Some may build their GUIs directly accessing the data
mining engine, as shown in Figure 3-10(a); others have the GUI using
the public API, as shown in Figure 3-10(b). Ideally, a tool should

Desktop GUI

Desktop GUI API

API

Data Mining Engine Data Mining Engine

Mining Object Repository Mining Object Repository

(a) (b)

Figure 3-10 Data mining tool architectures.

1 “C-level” refers to executive management (e.g., Chief Executive Officer, Chief

Information Officier).

2 A detailed discussion of JDM web services and architectural implications is

provided in Chapter 11.