IBM releases open analytics interface

Computerworld - IBM last week unveiled open-source technology designed so companies can apply analytics to call center logs, warranty claims, technician notes and other unstructured data.

IBM's Unstructured Information Management Architecture (UIMA) will help companies corral elusive unstructured data contained in documents, e-mails, Web sites and video for inclusion in their business intelligence and knowledge management efforts, users said. The UIMA framework provides standard interfaces to let corporate developers or IT vendors easily add unstructured information analytics to applications, IBM said.

In conjunction with IBM's announcement, 15 vendors, including SAS Institute Inc., Cognos Inc., ClearForest Corp. and Attensity Corp., last week pledged to support UIMA in their products. IBM also introduced a product called IBM WebSphere Information Integrator OmniFind Edition, which is based on UIMA.

Extending Investments

According to Tom Welke, vice president of North American operations at Whirlpool Corp., using UIMA as a standard interface will allow the Benton Harbor, Mich.-based company to extend its existing investments in text analytics to other areas.

Whirlpool has been using software from Attensity for the past three years to pluck unstructured data from the records generated from the 400,000 customer calls it logs every month in its call center. With UIMA, Welke said he hopes to be able to take that analysis and send it back to the call center in real time.

Regulatory DataCorp International LLC is deploying ClearForest's Text Analytics Suite, which is compliant with UIMA, to analyze thousands of sources of unstructured data daily, including media articles, government reports and Web sites, to add information to its Global Regulatory Information Database. The company was formed by 20 banks and financial services firms, such as Goldman Sachs & Co., Merrill Lynch & Co. and Citigroup Inc. It uses the database to help these companies lessen the risks of money laundering, fraud and terrorist financing, said Bob Catucci, president and CEO of New York-based Regulatory DataCorp.

Before using the ClearForest tools, company employees would manually search for information.

"If one of these terrorists was able to use your systems to transfer funds ... the government will be asking what you are doing to protect yourself," Catucci said. "You couldn't do this with an army of people reading articles."

Hadley Reynolds, an analyst at Delphi Group in Boston, said companies traditionally have been writing one-off interfaces to incorporate unstructured data into their analytic efforts.

UIMA is designed to let companies use the interface standard once to link unstructured data into multiple applications, he added.

The Push Into Unstructured Data IBM's Unstructured Information Management Architecture involved more than four years of research at IBM. IBM received support for UIMA from the Defense Department's Defense Advanced Research Projects Agency. IBM plans to make UIMA available through SourceForge before the end of the year; it's now available for free at the company's AlphaWorks Web site.

Read more about BI and Analytics in Computerworld's BI and Analytics Topic Center.