Is FAST enterprise search better than Google's?
- TAGS:data mining, documentum, enterprise, enterprise search, fast, Google, Google search appliance, Microsoft, pagerank, search engine, SharePoint
- IT TOPICS:Emerging Technology, Enterprise Apps, Privacy, Storage, Web Apps, Windows
That's what Microsoft Corp.'s Business Division president Jeffrey Raikes asserted Tuesday when he claimed that Google Inc.'s popularity-based search technology produces less accurate results for businesses than FAST's algorithms.
During a conference call after announcing plans to buy Norwegian enterprise search provider Fast Search & Transfer (FAST) for $1.2 billion, Microsoft's Raikes said that Internet search is all about "the number of hits to a particular site," referring to part of Google's vaunted PageRank search algorithm that favors documents that are clicked or viewed often by placing them higher in its results.
How popular is a document "doesn't matter inside a corporate intranet," Raikes continued, "where you have to look at the relevance algorithm to help the business user. That's what FAST has."
In his blog, independent search analyst Stephen Arnold appeared to agree with Raikes' characterization, though he did not come to a conclusion over whether FAST or Google was superior.
"'Behind the firewall search' has to be able to generate useful results when there aren’t indicators like the number of times a document is clicked on or viewed. As you may know, Google’s Web search system uses these cues to determine relevancy," Arnold wrote. "In an organization, a very important piece of information may have zero or very low accesses. In a patent matter, a 'behind the firewall search' system must be able to pinpoint that piece of information because it may be the difference between a successful legal resolution and a costly misstep."
Google, of course, sells an enterprise search appliance that competes with FAST and can search repositories such as Microsoft's content management platform, SharePoint and EMC Corp.'s Documentum.
Google also disagrees with Raikes' conclusion. In a discussion on how its algorithm has been tuned for its search appliance, Google says "some customers ask to what extent Google universal search technology relies on PageRank. The answer is: very little. PageRank is only one among more than a hundred factors that determine the relevancy of universal search. Even for consumer search – Google.com – PageRank is not the only factor."
Microsoft made similar comments at an IT conference in San Francisco last April, though a Googler was on stage at the time to give back as well as receive.
"We understand the needs of the high-end markets. There is some perception with a small segment of the market that we don't, but we have all the tools the market needs," said Nitin Mangtani, lead product manager for Google's search appliance business.
FAST CEO John Lervik also made an indirect threat to relational database and business intelligence vendors such as Oracle Corp. or Teradata Inc.
"Search is not just about handling unstructured data," he said. "It is also about searching numerical and structured data. We've been working on that for the last 7-8 years."
Lervik said FAST can also search "behavioral data" and help connect people to experts and colleagues, which sounded an awful lot to me like the sort of 'social network analysis' performed by the National Security Agency (NSA) among others as a form of sophisticated data mining.
So do you buy Microsoft and FAST's claims to superior technology? And how about FAST's possible sideways moves?
