Databases Reference
In-Depth Information
Value Decomposition (SVD). The algorithms are accessible via Java and PL/SQL APIs.
Other data mining capabilities available include text mining (providing document clus‐
tering and classification) and BLAST similarity searches leveraging the SVM algorithms
(common in genetic research).
Data mining applications can be custom-built using Oracle's Data Miner tool. Data
Miner is used to develop, test, and score the models. The data is usually prepared for
mining by binning, normalizing, and adjusting for missing values in the Oracle Data‐
base. Data Miner also provides the ability to define metadata, tune the generated Java
code, view generated XML files, and test application components.
Other Datatypes and Big Data
A number of other types of data are often stored in an Oracle Database. Among the
types of data are:
Multimedia and Images
The Multimedia feature set (once known as
inter
Media) opens up the possibilities
of including documents, audio, video, and some locator functions in the warehouse.
Of these, text retrieval (Oracle Text) is most commonly used in warehouses today.
However, the number of organizations storing other types of data, such as images,
is growing. Oracle first added DICOM imaging support, popular in medicine, to
Oracle Database 11
g
. Often, storage of these types of data is driven by a need to
provide remote users with access.
Spatial Data
The spatial locator capability in the Oracle Database enables retrieval of data based
on a geo-spatial location. Where distances and other geographic computations are
needed, the Spatial Option enables such applications to be built. An example of this
option's use for data warehousing is a marketing analysis application that deter‐
mines the viability of retail outlets at various locations.
XML
Oracle first added native XML datatype support to Oracle9
i
, along with XML and
SQL interchangeability for searching. Oracle provided key technology in the de‐
velopment of the XQuery standard, and began shipping a production version of
XQuery with Oracle Database 10
g
Release 2. XML database performance was greatly
improved in Oracle Database 11
g
through the introduction of binary XML.
Though most of this topic describes SQL as the primary way to access data, various
programming paradigms remain popular. As noted earlier in this chapter, MapReduce
is gaining in popularity for processing semi-structured data stored in Hadoop using
callouts from programming languages such as Java and Python. In some situations
where the organization would prefer to store all of the data in an Oracle Database and
