RealTime IT News

Enterprise Search: A Different Animal

SANTA CLARA -- While knowledge management tools are becoming critical for businesses, enterprise search is taking on a new value, Google officials said Wednesday.

Executives with the sultan of search were on hand at the KM World & Intranets trade show here to address the needs of businesses that need to store, retrieve and track digital information of all kinds. Enterprise search vendors made a strong presence at the show, set up alongside companies hawking digital rights and content management products, were other vendors reflecting the growing importance of search as an internal interface for corporate information.

Google admits it has a lot of competition in the knowledge management sector, but its hook is its Google Search Appliance, a hardware/software package that lets corporate users search through some 250 different document formats.

Matthew Glotzbach, business product manager for Google Search Appliance, evangelized that even for companies that use knowledge management tools that create taxonomies and meta tags for data, search is still key to finding corporate info.

"Search is the primary mechanism that users now use to find information," Glotzbach told internetnews.com. Just as people switched from browsing through Web directories to using search to hone in on just the Web pages they want, corporate users too tend to rely on search first. "People's expectations from the Web are carrying over to the enterprise."

Enterprise search is different from consumer Web search in two ways, he said: First, most information is not stored as Web pages on Web servers. Second, there's a different standard of relevance.

Corporate users need to find information stored in many different containers, including e-mail servers, desktops, enterprise application databases, content management systems, file systems, intranet sites and external Web sites. Enterprise search applications become a sort of virtual repository of corporate information, uniting these disparate sources behind one interface, said Mark Myers, vice president of strategic alliances for Convera. Today Convera announced it RetrievalWare 8.1, a search software platform that can be personalized.

Relevance must be defined differently when sifting through a company's documents. Often, relevance algorithms that work well for consumer Web search don't apply inside the enterprise. Convera's technology eschews popularity in favor of semantic analysis.

As Donald Taylor, senior director of Vivisimo's Life Science Solutions division pointed out, "One of the biggest challenges within the enterprise is that many of the documents stored are highly relevant. You might return 4,000 documents from an enterprise search and all 4000 could be relevant. The enterprise can typically overlook a lot of information as a result."

On Tuesday, Vivisimo launched Velocity, an enterprise search tool that combines dynamic clustering, search and meta-search.

And while Google doesn't discuss its ranking methods, Glotzbach said that the Search Appliance lets administrators fine-tune the different elements. "Popularity isn't one of them," he said.

But Entopia believes that popularity is essential to differentiating between the hundreds of company-generated documents that might refer to the same key words. The company introduced K-Bus 3, the latest release of its enterprise search software.

"We capture the context and user activity around documents," said David Hickman, vice president of product management for Entopia. "We know who wrote it, how many times it's been read, and whether anyone took the time to save, print or e-mail it."

Entopia ranks documents according to how often they're accessed and passed along. It also uses this information to identify experts and gives the content they generate a higher weight.

There's one way in which enterprise search is like consumer search: Most users don't know and don't want to learn how to refine searches to improve results. There are several strategies that help them refine searches and find what they're looking for.

Vivisimo organizes search results on the fly into folders containing related information. Users can click on folders that seem most relevant and drill down into them, dismissing those that are too far off. In a sense, the searcher is creating an individualized hierarchy while being prompted

"Key trends and themes emerge from the content, and they're hierarchical," Taylor said. " With our technology you can tease out ideas. It uncovers surprising relationships users may not have known existed."

Entopia takes a similar approach. Along with natural search results, it generates a list of the top ten concepts shared by documents relating to the search. Instead of reading through the results list, the searcher has the option of clicking on the concept that's the best match.

Convera lets users subscribe to alerts on topics of interest. It also lets those who subscribe publish their subscription lists so that others can use them. Neither of these processes is automated.

"It's self-selecting," said Convera's Myers. "You sort of know within your organization who the savvy people are."

Google Search Appliance allows administrators to select documents that will appear at the top of results for specific key words, not unlike the way Google.com returns a few paid listings at the top of natural search results. "It's almost like an internal ad, " Glotzbach said.

The next wave of enterprise search will be vertical tools that build in domain expertise, said Madanmohan Rao, author of the new book Knowledge Management Tools and Techniques.

"The best strategy," he said, "is to have a search engine that's finely tuned for a specific industry, such as pharmaceuticals, instead of a plain vanilla tool."