Microsoft's Plan To Bring Data Mining To Masses

Welcome Guest. | Log In| Register | Membership Benefits

May 25, 1999 (12:00 AM EDT)

Microsoft's Plan To Bring Data Mining To Masses

By Shawn Willett,

Microsoft Corp. wants to popularize data mining, an esoterictechnology that has only just begun to get off the ground in enterprises.

At its Tech Ed conference here, the company formally announced its OLE DB for Data Mining API, which aims to create a standard interface to retrieve database information for data-mining applications and a standard "model" for data-mining data.

Microsoft executives eventually plan to put data mining functionality into SQL Server.

"The end result for the customer is that this technology that used to be expensive and complicated now becomes part and parcel of their line of business applications, and customers get to take advantage of it," said Barry Goffe, lead product manager for SQL Server at Microsoft.

Data mining uses complex algorithms to search large amounts of data and find patterns, correlations and trends in data. IBM Corp. and smaller software firms traditionally have put a high price tag on the software and the consulting that is needed to decipher the data. Sometimes trained statisticians are needed to make sense of data patterns.

As a result the market has been relatively small.

Goffe said the standards should make it less expensive and allow ISVs to embed data-mining functionality into existing applications.

"A good example is CRM [customer relationship management]. A data-mining vendor can create a model that can identify who is a good customer. The CRM vendor can embed that model into the application," said Goffe.

By embedding certain models--for example fraud detection--into applications, the technology can be much more palatable to users, Goffe said. "Users don't even have to know it is there," he said.

Microsoft said six vendors--Angoss Software Corp., SAS Institute Inc., SPSS Inc., Silicon Graphics Inc., Data Sage Inc. and E.piphany--helped develop the specification. It is slated to become final by the end of the year.

Although it is not generally known, the Microsoft Research division has been working for years on data mining, said Goffe.

In addition, he said data-mining technology of some sort will be included in a future release of SQL Server.

Microsoft will leave plenty of room for third parties to take advantage of whatever data-mining functions are in SQL Server, Goffe said.

Already IBM Corp. has said it plans on putting statistical functions in its upcoming version of DB2 that it culled from its Intelligent Miner data mining product.