some of these standards and may have already implemented systems
and applications using them. Where possible, leverage existing data
mining standards to achieve a greater degree of interoperability and
avoid impedance mismatch between standards.
Tactically, we include the notion of import and export of objects
conforming to PMML and CWM, as well as provide a conscious
mapping of capabilities among data mining standards where possi-
ble and appropriate.
These are the strategic objectives for JDM. Next, we explore some
of the basic premises for standards.
Role of Standards
Although data mining in the form of artificial intelligence, machine
learning, and various statistical techniques has been around for
many decades, only within the past 5 years have data mining stan-
dards taken hold—a proof point for the maturing of the field and its
pervasiveness in the marketplace. This section explores the motiva-
tions for creating a standard and what standards enable for realizing
the Java Data Mining strategy.
Why Create a Standard?
Standards exist in nearly every aspect of life. In the physical world,
we have standards for phone jacks, electrical sockets, railroads, and
bathroom fixtures, to name a few. We feel and experience how these
standards make life much easier and products less expensive.
Consider electrical sockets: Anyone who has traveled to a foreign
country using a different electrical socket standard can readily under-
stand the benefits of standards.
In the nonphysical world, we have standards for Internet commu-
nication protocols, SQL, and programming languages. Those who
have used software and hardware products from multiple vendors
also appreciate the benefits of standards, or at least the problems
introduced by lack of standards.
The Java programming language is a specific case of a standard
that benefits its user community. In particular, the Java Data Mining
standard provides a common framework for exploring and develop-
ing applications using data mining. Developers are able to learn a