Commerce-tested search tool bows

CAMBRIDGE, MASS. – Software developer Endeca, which is boosting sales for companies such as Barnes & Noble by making it easier for buyers to find what they’re looking for online, is bringing its guided navigation search technology to corporate intranets.

The company will announce ProFind, an enterprise search product based on the Navigation Engine that powers Endeca’s InFront e-commerce software, at this week’s Gartner Symposium/ITxpo in Orlando.

While InFront primarily is used with commerce catalogs, ProFind can be used to manage structured and unstructured content within a company, letting business users search everything from documents, PDFs and enterprise resource planning data.

The software works by extracting data from corporate databases, and from unstructured content sources such as e-mail messages and Web pages, and indexing it in the Navigation Engine.

The information then is searched using what Endeca calls guided navigation. Guided navigation goes a step beyond traditional searches by providing users with a list of subheads to help narrow the search.

The subheads listed on the search screen change dynamically as the search proceeds, weeding out dead ends.

Guided navigation helps guard against what the company calls the million or none problem with traditional searches: underspecified queries based on a few words that return long lists of options, and overly specific queries that turn up nothing.

“We turn that paradigm on its head,” says Paul Whitelam, product manager at Endeca.

“We show the user the choices that they can make to refine a search result. When you get a result set returned by Endeca, that’s really the beginning of your exploration,” he adds.

For example, an engineer searching an automotive database for the specification for paint used on Ford’s bumpers starts by typing the word “bumper” into the search box. That returns about 500 results from 40,000 available documents.

Endeca also returns a number of subheads that the Endeca Navigation Engine dynamically generates from the metadata in the current set of documents.

Clicking on the “sources” option, shows sources of relevant documents. The engineer clicks on Ford, reducing the number of results to about 190. As additional subheads are clicked, the list of results narrows.

Unlike taxonomy-based search tools, Endeca updates navigation options at each click so users aren’t forced down a path defined by a taxonomist.This all provides context to the search, letting users intelligently explore available information, even finding data they might not have known existed, Whitelam says.

Endeca executives say that because Endeca indexes metadata extracted from databases, the software is scalable and doesn’t demand high-end hardware that some other search vendors require.

Information Handling Services Engineering, a provider of technical standards, specifications, logistics and parts information in Denver, scrapped “four Sun 880-class machines with a lot of redundant disk space” when it moved from Convera’s search product to Endeca earlier this year, says Paul Magin, vice president of product development at IHS.

Reducing hardware costs

IHS replaced the Sun servers with three x86 Solaris boxes, cutting annual hardware costs by about $100,000, Magin says. That should slice annual hardware and administration costs in half, he says.

IHS also found it was easier to load content into the Endeca system. Five applications used to manage more than a million documents were migrated from Convera and other search systems in seven weeks. That compares with three months to build a single application with the old tools, Magin says.

At the same time, users have experienced a fourfold increase in the response time for their searches. “The performance was astounding,” he says.

Updating content also is easier, Magin adds.

“In Endeca we can quickly model an extract of our main collection and index it in a few hours,” he says.

Analysts say information overload is costing millions of dollars in lost productivity because of time wasted searching for information. IDC estimates that knowledge workers spend 15% to 33% of their time looking for information and are unsuccessful about half the time.

With most search tools, users end up frustrated because they’re not asking questions correctly and not getting the answers they need, says Sue Feldman, research vice president of content management and retrieval software at IDC. Guided navigation improves on a search by providing choices to users to help them search smarter, she says.

“So somebody can say, ‘Oh, I can go here or I can look there. I didn’t realize that.’ Avoiding dead-ends is also important so you don’t send people away discouraged. Endeca does all of those things,” she says.

ProFind is available immediately starting at about $100,000 per license, although pricing is dependent on a number of variables, including number of users and feature.