RealTime IT News

Search With the Human Touch

A new crop of search engines bases relevance on the human mind instead of algorithms.

On Thursday, PreFound.com said it would begin to share ad revenue with active users of its site.

PreFound is a community-oriented search engine that lets people search, tag and share their searches with others; it launched in January 2006. Members can create their own personal pages containing organized groups of links sorted by topic and sub-topic.

Experts known as "Featured Finders" who take the time to gather the most relevant and recent information on specific topics will receive 100 percent of the Google AdSense revenue that their individual pages generate.

Featured Finders will be responsible for finding, tagging, and sharing organized sets of links and annotations on their expert topic areas that they have found on the Web. Searchers will be able to visit each expert's page and view the Featured Finder's PreFound selections.

"Our plan is to reward experts in their fields along with regular users who've proven they can share useful and popular material," said Steve Mansfield, CEO of PreFound.com and its parent company, iLOR, in a statement.

According to Mansfield, paying people will help ramp up the site's content, while reducing the visibility of spammy or slopping pages.

Experts interested in participating in the PreFound.com Featured Finders program can apply at www.profound.com.

Other sites too are looking for the human touch in improving search results.

Dumbfind, a two-year-old search provider, also plans to offer an online community where users can submit and tag content to help determine search relevancy and discover new sites. Dumbfind combines tag search with keyword search to better refine results.

Dumbfind automatically tags all the content in its database, groups them, determines which are most relevant, and displays them in clusters. Founder Chris Seline said the concepts of tagging and "folksonomies," wherein users create their own categories, are a good jumping off point.

"We have a large seed database of tags," he said, "but it could benefit from user input. There may be certain ways people might describe things that our technology might not capture."

Improving results could also improve ad revenue. PreFound's Mansfield pointed out that users of his site's topic-specific pages are more easily targeted by advertisers.

Eurekster recently launched Swicki, a search/wiki combo that lets Web publishers and bloggers offer topic-centric searches designed to give their communities of users more relevant results.

Publishers can focus and train their Swickis by typing in keywords and relevant URLS.

Then, Swicki technology automatically learns from search behavior on the publisher's site, constantly refining search results in response to what site users clicked on. For example, a traditional Web search for the word "labor" might return results focused on childbirth, labor legislation and unions. On a site catering to pregnant women, searchers consistently clicking on links related to childbirth would eventually increase the relevance ranking of such links, so that only they would be shown to users.

In a statement, Eurekster CEO Steven Marder said, "Currently, publishers lose traffic to generic search engines because they don't offer their users a Web search with a differentiated or specialized value-add that retains them. Unique search results not only help build user loyalty, but also lead to greater search-driven advertising revenue, he said.

The idea of human editors is as old as Web search itself, of course. The original Yahoo was a simple, human-edited list of links to sites. About.com, founded in 1996 as The Mining Co., paid subject-expert Guides to create channels. About.com was sold to the New York Times Company in February 2005.