Simulating Google’s secret sauce
- By Linda L. Briggs
Could you use your enterprise search product to improve how well your externally
facing Web pages might do in a Google or Yahoo! search?
Well, yes. That’s what Jeff Watts is doing in his capacity as search
and community manager for National Instruments, which makes software and hardware
for scientists and engineers. Part of his job is to drive traffic, including
potential customers and users with technical questions to the company’s
Web site. By emulating Google’s search capabilities internally,
the company can adjust external-facing Web pages to improve search performance.
Without knowing the secrets behind Google’s proprietary search system,
but by using the company’s enterprise search solution, Fast Search Enterprise
Search Platform, Watts and his colleagues were able to better understand how
Google, Yahoo! and other search engines might measure the relevancy of a page.
“To emulate Google is impossible,” Watts says. “But
to create something like Google is relatively easy within Fast Search.”
He runs the improvised search engine internally against Web content before it
goes live. “You can…do things like apply a heavier weight to the
number of inbound links for a particular page—something that very grossly
emulates Google’s PageRank algorithm.”
The process is undergoing more testing, Watts says. Already, the company has
tweaked pages to make a vast improvement in its online technical support efficiency.
“Other groups [in the company] are definitely looking at this,”
The project was possible partly because Fast Seach has an open structure
that allows customers like Watts to look at, and even change, the relevancy
model—the complex set of algorithms that determines the ranking of items
found in a search.
Back to Feature: Search
Engines Look for Answers Inside the Enterprise