Vendor Implementations - Java Data Mining: Strategy, Standard, and Practice - page 420

Java Application

Java Application

JDM API

JDM API

DME

DME

MOR

MOR

(a) (b)

Figure 16-3 KJDM choices of architectures.

Figure 16-3(a) represents both the Java application and the JDM
implementation in the same process. Figure 16-3(b) represents the
Java application and the DME implementation in two different pro-
cesses, which can be on the same machine or on different machines.
For each case, the software architect has the choice of using an MOR
accessible only on the machine, or an MOR residing on another
environment: In this latter case, the data contained in the MOR will
be accessed through open database connectivity (ODBC) by the DME
itself. Figure 16-3(b) highlights a three-tier architecture.

16.2.4 KXEN JDM Capabilities

KXEN JDM (KJDM) supports all the JDM 1.1 mining functions:
classification, regression, attribute importance, association , and clustering .
There are some key features of KXEN JDM implementations:

• KJDM does not implement any specific algorithm: Java pro-
grammers can stay at the mining function level without any