News

IBM's Data Mining Tool Comes of Age

• By John K. Waters
• April 2, 2001

SAN JOSE, CA—Now that data mining has emerged as a key competitive technology, sophisticated tools that allow users to browse, merge, manipulate, and analyze vast stores of raw enterprise data, and display the results in a wide range of useful forms and formats, are maturing fast. Among the most interesting of the current class to hit puberty is IBM's Intelligent Miner Scoring 7.1, which was just released.

Big Blue's new product has been implemented as an extension to its DB2 Universal Database. Intelligent Miner Scoring works directly from the relational DB, speeding up the data-mining process. Because it is an extender to DB2, it can easily read, not only DB2 data, but data running on competitive databases as well. For example, the product is compatible with Oracle databases.

Bringing the data mining functions into the DB engine reduces the amount of instructions the computer has to execute to respond to a user. The result, says Jeff Jones, senior program manager in IBM's Data Management Solutions Group, is real-time, interactive data mining.

"By taking the mining capabilities and building them as an extension to DB2, the notion of real-time data mining—data mining while you're talking on the telephone with a customer—is feasible," Jones says. "We believe that we are the first to implement integrated data mining into the database engine."

In line with the industry's efforts to enhance the exchange of information on the Web with tools like Extensible Markup Language (XML), Intelligent Miner Scoring also supports the industry standard for predictive modeling, PMML (Predictive Modeling Markup Language). PMML is an XML-based language that provides applications a vendor-independent method of defining predictive models to eliminate proprietary incompatibilities. IBM is among the first to support the PMML standard in an implementation of a mining tool.