RealTime IT News

Enterprise Search is About The 'Folks'

NEW YORK -- Forget about the magic bullets of either taxonomies or user-generated tags.

Information architects will have to roll up their sleeves if users are to ever believe they can search corporate intranets with any sense of confidence, said two keynote speakers at the Enterprise Search Forum here today.

Peter Morville, an information architect and author, said that search engines only deliver value to an organization by delivering meaningful results and that neither rigid taxonomies nor more fluid user-influenced tags can do the work.

Martin White, a fellow consultant based in the U.K., agreed that information architects will have to continue refining search engines for corporate intranets.

The first place to start, according to Morville, is with the search results page, as it plays a vital role in making the search engine credible.

User logs show that search results pages get the lion's share of traffic on corporate intranets, second only to the homepage.

Yet, according to Morville, companies pay virtually no time or effort to make this page more usable.

"The results interface is one of the biggest opportunities for content sites to improve the user experience," he said.

This is important, he added, because enterprise search has to provide value, and it cannot do so unless users trust that the search engine is reliable.

Search result pages also speak to how easy it is to identify and navigate content, another factor he identifies as vital in establishing credibility.

"Visual design and information architecture have a big impact on advancing the goals of the business," he said.

Morville noted that Wikipedia enjoys a high degree of credibility because of its design and the methodology used to structure content.

Although the content itself is user-created -- and therefore susceptible to providing inaccurate information -- the Web site's traditional user interface "enables 'findability' and provides credibility and authority," he said.

Navigation in Wikipedia features what information architects refer to as an "inverted L."

Morville called this "a fixed, very traditional approach. The user can't change that," he said. "That's why we trust the content in the body of the page."

In the end, however, search results can only be as good as the principle for the way the content is organized in a database.

Currently, the world of search is divided into two warring camps -- those who believe that content should be organized into rigid taxonomies, and those who believe that users themselves are best suited to tagging content.

Rather than choosing between them, said Morville, information architects should use taxonomies to establish an infrastructure and governance rules for the search engine, and use more fluid "folksonomies" to keep up with changes.

He described the overall structure of an ideal content management system in terms of layers, with rigid taxonomies on the bottom and user-based tagging on top.

"Controlled vocabularies and taxonomies form some of these lower layers, and provide some sense of stability. Folksonomy is more fast moving, and allows us to learn and adapt very quickly. We can draw lessons from the top to inform the lower levels," he said.

However, several observers took issue with the usefulness of taxonomies.

White argued that taxonomies have a hard time keeping up with changes that occur in the business.

He suggested that someone searching for documents relevant to a particular project would get incomplete results if the name of the project changed.

"Is the search engine supposed to guess?" he asked.

Sue Feldman, an analyst with IDC, a research firm based in Framingham, Mass., agreed that taxonomies are too rigid for rapidly evolving contexts.

"Taxonomies are backward-looking," she told internetnews.com. If you depend on them to organize information, searches "are going to miss the new stuff."