Searching for IBM

The market for enterprise search might seem like a three-horse race between Google , Microsoft and Yahoo , but another contender has the sector looking over its shoulder: IBM .

Researchers in IBM’s information management systems unit have been discussing intelligence systems that are highly intuitive, with new software systems that go beyond traditional search by pulling subtext from stored data. This includes structured information in databases as well as unstructured data like e-mail or video files.

For example, the Armonk, N.Y., company recently took the covers off OmniFind, an enterprise search engine for file systems, content repositories, databases, collaboration systems, applications and intranets. OmniFind integrates with any portal or content management system.

Used in IBM’s DB2 Information Integrator middleware, OmniFind is designed to spearhead the next evolution of enterprise search. This type of search is different from traditional Web-based approaches in that it requires more precise techniques to determine document relevancy, taking into account different security models, data sources and file types.

Currently, search techniques include simple keyword matches on desktops and Web-based search conducted by the likes of Google that lets users retrieve data from several Web sites. But Internet search techniques such as page ranking are not optimized for an enterprise environment where documents are not as interlinked to each other.

OmniFind is part of IBM’s grand plan for unstructured information management architecture (UIMA), a technology approach for culling the underlying meaning from stored data regardless of where it lies in a computer system.

IBM also hopes to make waves in natural language processing with systems like the Practical Intelligent Question Answering Technology (PIQUANT).

PIQUANT analyzes the semantic structure of a passage, culling information that wasn’t overtly present on a database or file system. This enables users to find answers to specific questions even if the keywords they use do not exist in the article they’re searching. Other natural language processing functions work to translate sites in multiple languages into one desired language.

Another system gaining a lot of attention is IBM’s WebFountain. This platform uses the red-hot Web services approach to distributed computing to conduct highly specialized searches. WebFountain converts the disparate ways information is presented online into a uniform format that can be analyzed.

The platform allows access to databases, applications and other repositories, enabling the creation of a set of hosted Web services containing information that drives end-user applications. Analytical components can be authored remotely by partners using a collection of Web service APIs .

IBM believes the text analytics capabilities of WebFountain will lead the way in a market that could be worth billions of dollars, according to experts. John Battelle, a search expert and co-founder of Wired magazine, agreed the potential of WebFountain is great.

Battelle also said on his blog that despite their obvious difference in target audiences — Google focuses on consumers while IBM caters to businesses — “WebFountain and Google are at least kissing cousins,” noting that they were inspired by an earlier concept of a system that would count inbound and outbound links to identify central sites in a community.

Battelle said the two companies are racing toward a middle where they may well meet because they are both trying to solve the problems of providing more reliable search.

It boils down to more intuitive search. Google will find you the answer to questions so long as those key or subject words appear in an article. Systems like OmniFind, PIQUANT and WebFountain were designed to go deeper.

Despite the promise, IBM has ground to gain among the Google’s, Microsoft’s and Yahoo’s in the search space. As further proof of the serious attention IBM is paying to search technology, the company has nailed down a number of patents in the field.

But all players have their eyes on enterprise search as well, where IBM enjoys a commanding presence with its installed base of hardware and database software. Big Blue already sells
DB2 Information Integrator software that lets users pull information from multiple data types from one location.

Google’s desktop
search beta
is seen as one facet of the search player’s enterprise search plans. Yahoo also trotted
new search capabilities in October.

Microsoft is turning heads with talk of its WinFS file system, although it will not appear
as part of the company’s next operating system, Longhorn, and may take five years to arrive.

Still, it is difficult to bet against the mighty intellectual and fiscal capital of Big Blue.

News Around the Web