Strategic Objective 8: Support requirements of real,
A major impetus to businesses using data mining is how to derive
value or operationalize their data mining results in business applica-
tions and processes. Deploying data mining results can involve mov-
ing the data mining models from the lab into the field, or publishing
the results in reports or through dashboards. The complexity of
applications often requires the ability to transport data mining
objects and related information between machines.
Sophisticated graphical user interfaces (GUIs) for performing data
mining can be feature rich with algorithms, transformations, and
support for model building, test, and scoring. However, GUIs are
often not enough to deploy data mining solutions in applications and
throughout a business.
Tactically, we enable applications to exchange objects using export
and import interfaces, to save mining objects to files, and to share set-
tings and results, including models, between vendor implementa-
tions. The ability to transport data mining objects from one
environment to another is often key in a distributed application envi-
ronment. For example, where models are built in one system but
applied to data in another, an XML representation of mining objects
is highly valuable. Adopting a common XML Schema representation
for JDM objects facilitates interchange among vendors, as discussed
in Chapter 10.
Strategic Objective 9: Appeal to vendors and architects
in other development domains
Implementing and conforming to a standard can be a major
undertaking. If the data mining engine (DME) must be modified or
enhanced to support a standard interface, vendors are more inclined
to adopt that standard if it can reach a broader audience.
Tactically, we enable data mining in a Service Oriented Architec-
ture (SOA) by defining a web services interface that maps closely
with the Java API. This allows application designers and developers
to freely move between Java and web services, especially within the
same vendor implementation.
Strategic Objective 10: Leverage other data mining standards
Much effort has gone into standardizing various aspects of data
mining. Some vendors and consumers are already familiar with