Trends in Data Mining and Knowledge Discovery - Advanced Techniques in Knowledge Discovery and Data Mining - page 14

Services Description Language ( WSDL), which is the XML-based format for
describing network services [73]. At the same time, because of benefits like
reaching new customers, expanding offerings, and market reach, it is almost
certain that service providers will register their services using UDDI.
From the DMKD point of view, services can be DM tools that are published
as online services. DM toolboxes can be implemented as clients that can use those
services [45]. The DM toolbox would then check the availability of the online DM
tools using UDDI and invoke the ones that can provide meaningful results for the
currently processed data. The DM tools (services) then would take the data
provided by the DM toolbox, process it, and return results to the toolbox. Using
this protocol, a DM toolbox can dynamically access and use several DM tools,
which process data and generate results. The toolbox would collect the results,
process them, present them to the user, and finally store them in the knowledge
base. This simple mechanism, powered by dynamic online DM tools, can be used
to build flexible and widely applicable DM toolboxes.
These technologies will certainly help in semiautomating the DMKD process.
XML can be used to store data and PMML to store data models. SOAP and XML-
RPC can be used for platform-independent communication between different DM
applications, and UDDI can be used to find DM services offered by DM
companies. A more detailed description of how to incorporate the technologies
into the DMKD process is given later. A big advantage of these technologies is
that they are open source and thus can be freely downloaded and used.
1.3.6. OLAP
OLAP (online analytical processing) is a relatively old DMKD technology. Its
main purpose is to provide users with multidimensional views of aggregate data
for quick access to the needed information for further analysis. OLAP gives fast,
consistent, interactive access to a variety of views of any information. OLAP and
data warehouses (DW) are complementary technologies. A DW stores and
manages data whereas OLAP transforms the data into possibly strategic
information. OLAP services range from basic navigation and browsing (called
slice and dice) data, to analyses such as time series processing. OLAP gives the
user some decision-making power. The most common applications of OLAP are
marketing, promotions, customer analysis, sales forecasting, and market and
customer segmentation. OLAP has the following characteristics:
z multidimensional views , which help in analytical processing of the data
through flexible access to information. Users can analyze data in any
dimension and at any level of aggregation.
z time intelligence , which means that OLAP systems can deal with the
sequential nature of time. The notion of time should be built as an integral
feature to any analytical package.
z complex calculations , which give the user a tool to, for instance perform share
calculations (percentage of the total), allocations (which use hierarchies from
a top-down perspective); they use trend algorithms such as moving averages
and percentage growth.