Oracle Database 11g is launched: New Oracle Data Mining (ODM) 11g features

Oracle Database 11g is official launched today with many new useful features. I am excited to launch my blog on the same day.

There are many new data mining features in 11g that will greatly ease and enhance the data mining solution development for the users.

One of my favorite 11g new feature in Oracle Data Mining (ODM) is Automated Data Preparation (ADP) that enables automation of many complex data mining specific transformations done transparently as part of the model build process. This feature is further extended to support embedding user specific business transformations with the mining model that will greatly simplify the model deployment to produce useful results for the applications.

My next favorite feature is Mining Model in 11g is one of the database schema object (like Tables etc.). With this feature, mining models inherit powerful database security features, such as the ability to grant data mining specific system privileges or object privileges. For example, a DBA can grant only the analyst user(s) ability to create models and application users to be able to use those models for applying and/or viewing. Similarly, a user can grant object privileges on specific mining models that s/he owns. For example, an analyst can give select privilege for only some models that are production ready in his/her schema to the application users.

In this release new Generalized Linear Models algorithms are added to ODM for both classification and regression functions.

Another useful feature in this release is data mining models meta-data integration with the core database dictionary. This enables applications to easily query the information about the models such as settings, attributes, data preparations used by the model.

If you are doing text mining or using or wanted to mine multiple nested tables this new feature will be interesting for you. In 11g nested tables attributes are scoped by the column. This will greatly ease unique attribute name issues that are typically encountered in text mining where user is mining more than one text column in the previous releases.

All these new features are enabled for Java and PL/SQL Data Mining API's provided with the ODM option. There are some Oracle Java Data Mining API specific new features, that enable applications to build simple work-flows using the new task dependency feature and ability to set overwrite flag in the task object.

For more details on ODM 11g new features keep an eye on ODM product website http://www.oracle.com/technology/products/bi/odm/index.html and also there is a new RSS feed available in this website that will enable you to get the information about ODM into your personal pages. I will post links in this blog if I find any new articles on ODM 11g.