Trends in Data Mining and Knowledge Discovery - Advanced Techniques in Knowledge Discovery and Data Mining - page 13

RPC, and it is a superset of XML-RPC, but they are not compatible.
From the DMKD point of view, both XML-RPC and SOAP can be used as
protocols to communicate between DM tools to create DM toolboxes. Such
toolboxes would use multiple DM tools, choosing ones that are suitable to work
with the supplied data and provide the user with combined results without the
necessity of running the data separately using all chosen DM tools [45]. Using
these protocols, the DM toolbox can access the DM tools over the Internet; as a
result distributed and user-customized toolboxes can be easily built.
1.3.4. PMML
PMML (Predictive Model Markup Language) is an XML-based language used to
define predictive data models and share them between compliant applications [56].
PMML was designed by the Data Mining Group (DMG) [29]. DMG is an
independent vendor-led group that develops data mining standards; its members
include IBM, Oracle, SPSS Inc., Angoss, and MineIt Software Ltd.. PMML is
supported by products from IBM, Oracle, SPSS, NCR, Magnify, Angoss, and
other companies.
PMML defines the vendor-independent method for defining models. It
removes the issues of incompatibility between applications and proprietary
formats. This, in turn, enables exchanging models between applications. For
example, it allows users to generate data models using one vendor application and
then to use another vendor application to analyze, still another to evaluate the
models, and yet another to visualize the model. This is yet another very important
element that would enable building DM toolboxes. Previous solutions to the
problem of sharing data models were incorporated into custom-built systems, and
thus exchange of models with an application outside of the system was virtually
impossible.
The PMML currently supports the following DM models: decision trees,
naive Bayes models, regression models, sequence and association rules, neural
networks, and center- and distribution-based clustering algorithms [29]. The
PMML describes the models using eight modules: header, data schema, DM
schema, predictive model schema, definition for predictive models, definition for
ensemble of models, rules for selecting and combining models and ensembles of
models, and rules for exception handling [36]. The PMML supports not only
several DM models but also the ensemble of models and mechanisms for selecting
and combining the models.
1.3.5. UDDI
Universal Description Discovery and Integration (UDDI) is a platform-
independent framework for describing, discovering, and integrating services using
the Internet and operational registry [71]. The framework uses XML, SOAP,
HTTP, and Domain Name System (DNS) protocols. Currently more than 220
companies use the UDDI. The UDDI can help solve problems like finding the
correct service among millions available or interfacing with a service using Web