IBM Turns Over Search Project To Open Source Community
"Developers can start working with the code," said IBM's Mark Andrews in an interview. "Before UIMA, there was no easy way to (collaborate on) search." The firm plans to move the entire project to a full open source community development model before the end of the year.
IBM said it is publishing UIMA source code to open source development site SourceForge.net.
UIMA, which pulls together unstructured content from a variety of sources, has been spearheaded by IBM, but Andrews noted that the collaborative search solution has been gathering momentum in recent months, picking up partners and developers.
"In and of itself UIMA does not extract knowledge," said Andrews. "It provides interfaces, provides a common framework that enables you to plug and play various technologies to extract knowledge and incorporate it into enterprise business applications and intelligence applications."
As the firm released the UIMA source code, it pointed to early uses of the technology. Andrews noted that the Mayo clinic has used IBM's unstructured text processing capability with the UIMA framework to extract knowledge from its 20 million clinical notes.Another IBM UIMA program partnering with Memorial Sloan-Kettering Cancer Center is building an online data warehouse that conforms with HIPAA compliance requirements. Andrews said IBM's UIMA text analytics solution serves as an aid to search Memorial Sloan-Kettering's pathology reports.
Andrews, who is a content specialist in the Software Group's Strategy and Business Development at IBM, said UIMA technology is embedded in different IBM products including WebSphere Information Integrator OmniFind Edition.
Does UIMA represent competition to established search engines such as Google, Yahoo and Microsoft, Andrews was asked?
"We believe it will eventually help the Googles and the Yahoos of this world," he said. "People are finding that typical keyword search and page ranking aren't enough. We're driving UIMA as a broad industry standard. There are 25 to 30 companies, a bunch of open source (organizations) and universities developing it."
UIMA was first unveiled in December of 2004 as a collaboration by several corporate and open source partners. The Defense Advanced Research Projects Agency (DARPA) is using UIMA in intelligence applications to analyze huge volumes of speech and text in different languages.
Other companies at the vanguard of UIMA that are making developmental work widely available include ClearForest, Cognos, Factiva, and Nstein.