OmniFind: IBM’s Search Work Horse

March 17, 2008

This weekend (March 15 and 16, 2008), I engaged in a telephone call and a quick email exchange about IBM’s “behind the firewall” search strategy. I did include IBM in my new study Beyond Search for The Gilbane Group, but I narrowed my remarks to the search work horse, IBM OmniFind. According to my sources, OmniFind makes use of Lucene, an open source search system. I wrote about Lucene in this Web log in January 2008.

In the last week, a barrage of information appeared in my RSS newsreaders. The headlines were compelling because IBM relies on skilled word smiths; for example, “IBM Upgrades Enterprise Search Software“. Catchy.

What’s New?

What are the enhancements to OmniFind Enterprise Edtion 8.5? Let’s take a quick look at what’s new:

  • An improved user interface. IBM has introduced a “dashboard” called Top Results Analysis. The dashboard provides a single point for Lotus Notes / Domino information. The dashboard supports Japanese, Korean, and Chinese double byte encoding.
  • Beefier Connectors to Lotus Notes / Domino and Lotus Quickr* service so OmniFind users have access to IBM’s behind-the-firewall collaboration system which IBM seems to be suggesting is now a “social networking function”. *[Quickr is a newish team collaboration component.]
  • Enhanced support for the FileNet P8 content management system.
  • Support for ProAct, a business intelligence and text mining application that supports the IBM “standard” known as UIMA (Unstructured Informatoin Management Analysis). A simplified explanation of UIMA is that UIMA lays out rules so it is easier for IBM partners, customers, and third-party developers to “hook into” the IBM WebSphere and OmniFind frameworks.

My calls to IBM went unanswered. No surprise there. I dug through my open source files and came up with some information that may presage what the “dashboard” interface will permit. IBM has technology described as the “WebSphere Portlet Factory”. The idea is that “portlets” can be used in an interface. In my files, I located these examples of “portlets”. The OmniFind interface makes use of this technology to provide data about search results graphically, one-click access to collaboration, and similar “beyond search” functionality. The graphic displays appear to make use of Cognos technology, acquired in late 2007.


This illustration combines screen grabs of various IBM portlet functions. Each “portlet” comes from different IBM open source materials.

If your blood runs IBM blue, you will recall that IBM offers a mainframe search system, hyped WebFountain, and invested significantly in semantic search. One IBM wizard defected to Google and “invented” semantic technology which Google has disclosed in its patent applications for its Semantic Web. You may also recall that whatever IBM “sells” is not an application that you load on your server and head home early. IBM’s approach is to make available components, services, and hardware. You can dip into the IBM virtuous circle for comparatively modest sums. For instance, the iPhrase technology for call center self-service support starts at less than $5,000. Scaling to enterprise-class platforms, not surprisingly, requires a more substantial investment. IBM — like Microsoft, Oracle, and SAP — deploy systems that routinely hit seven figures and often soar higher.


The IBM release of version 8.5 of OmniFind triggers several thoughts. Let me highlight four of them. If you disagree, let me know.

First, OmniFind 8.5 affirms IBM as the digital Julia Child. Mrs. Child — a self-taught French chef and TV personality — crafted recipes any American housewife with patience could convert to “real” French cooking. OmniFind is not an application. It is a group of components, servers, and functions. To make an IBM solution, you combine the ingredients. There, you have it.

Second, I have a tough time figuring out where the IBM-developed technology goes. The defection of Dr. Ramanathan Guha to Google and the five PSE patent applications struck me as important. But IBM has quite a bit of innovation that seems to fade when the company acquires another firm. A good example is the iPhrase technology. IBM uses this technology, but it did not make use of the semantic inventions from the unit where Dr. Guha worked. I find this interesting?

Third, each year I vow to track down a person who is in the IBM search unit. This year I tried to contact Aaron Brown, whom I heard was the head honcho of IBM’s Information Management Group’s search unit. No joy. Each year I remain puzzled over the various versions of WebSphere, the differences between OmniFind Discovery and plain vanilla OmniFind, the search functionality of DB2, the role of SearchManager/370 (a mainframe search solution), the WebFountain text services, the free Yahoo edition of OmniFind, and the search oddments in Lotus Notes, among others.

In closing, I will track down OmniFind 8.5, load it on one of my aging NetFinity 5500s, and give the system a shake down run. You probably thought that I was anti-IBM. Wrong. I’m a believer in the firm’s hardware, and I’ve done a number of projects with WebSphere, “portlets”, DB2, and various other cogs in the IBM machine.

I am frustrated. I can’t figure out the product line up, how the pieces fit together, and how to navigate among the large number of options and combinations. If a fan of IBM can’t figure out OmniFind and its enhancements, potential customers have to crack the code.

Stephen Arnold, March 17, 2008


Comments are closed.

  • Archives

  • Recent Posts

  • Meta