updated 2/9/2006 6:35:04 PM ET 2006-02-09T23:35:04

Despite powerful search engines, the vast and rising sea of information on the Internet often makes users feel like they're stuck alone in a rowboat when it comes to fishing out exactly what they need.

Even after someone finds an answer, the next person with the same question must start all over again. Sometimes, information is locked away inside a photo or other media file that's largely inaccessible to search engines that scour the Internet for text.

The problem with corralling results from today's search engines was a major theme at this week's DEMO tech conference, where nearly 70 companies were given six minutes each to show off their up-and-coming products.

"There's a belief that whatever it is I'm looking for is out there, but I have a really difficult time finding it," said Chris Shipley, executive producer of the elite show. "Search algorithms alone are falling short in being able to provide real context around information."

A startup named Plum unveiled a service that lets users group Web pages, e-mail, music, pictures and files from their desktop computers into online collections that can be kept private or made public for others to find.

Web pages can be added to a Plum collection by clicking on a browser toolbar button, which calls up a box for choosing a picture, adding a description and creating an identifying tag. Content on the PC, such as a photo, can be added by clicking in a small program that imports files from a variety of applications.

Public collections are analyzed by the service and others that might be of interest are displayed, said Plum co-founder Hans Peter Brondmo. Eventually, collection pages will contain relevant ads.

"It gets more and more interesting with the more users you have in it," he said. "We take a look at all this information that others are putting into their collections and we match your information with it."

The secret sauce is an algorithm that analyzes and tries to match content — not unlike how a shopping site makes product recommendations based on previous purchases.

A Plum page created by Brondmo's wife, for instance, displayed links to collections on party planning and cooking gadgets. Both topics are mentioned in her pages.

Designed to be totally open and easily editable, Plum collections can be e-mailed to other users and even syndicated as an online "RSS" feed. If you subscribe to someone else's collection, you can immediately see when content is modified or added.

Plum is undergoing private testing, and the company has not yet announced a date for public availability.

It's hardly the only company seeking to tap the collective intelligence of computer users. Others — such as the site del.icio.us, recently acquired by Yahoo Inc. — offer tools for aggregating and sharing bookmarks and other content online, though not quite to the extent that Plum promises.

"Our goal at Plum is to make collecting and sharing anything online as quick and easy as using e-mail," Brondmo said.

Another DEMO presenter, Kaboodle, also is aiming to make online searching a social activity — but with a focus on shopping. It's currently available as a free public test.

"The trouble with online shopping today is there are too many choices," said Manish Chandra, Kaboodle's founder and CEO. "You have to visit multiple sites before you make any decision. It's hard to get all that information in one place."

By clicking a button on a Web browser's toolbar, a user can add a page of interest to Kaboodle. On e-commerce sites, Kaboodle grabs the price, a picture and description of an item and stores it on the user's Kaboodle page.

If the creator makes the pages publicly available, they can be found by others either through Kaboodle's search box or other search engines. Users also can post notes and comments to their Kaboodle pages.

"Kaboodle.com is a great example of people-powered search, allowing the work of one to be leveraged for the benefits of others," Shipley said.

Another company at DEMO, Riya, is trying to tackle two headaches: the difficulty of sharing information and the fact that photos can't be searched very easily unless they've been properly tagged with text descriptions.

The free service, which is expected to open for public testing this month, relies on face and text recognition to look inside the photo. It keeps a copy on its servers, which the user and others can search by typing in a name of someone in a picture.

"There are a lot of online services for printing, sharing and organizing your photos but not one that automatically searches them," said Riya co-founder Manjul Shah.

After users upload their pictures, the system scans the images for faces. Those that aren't identified — likely all of them at first — are displayed. The user can then manually identify a person by clicking a box that appears around the mug and entering the name.

It also can scour a user's address books for contacts who have used Riya to identify faces. If any are found, their previous work in training the system to recognize faces is automatically inherited by the new user.

During a demonstration of Riya's search capability, Shah typed in his son's name and the system returned hundreds of hits, with very few mistakes. In one case, it found the boy's face even though it was in a picture hanging in the background of a photo.

From the brief presentation at DEMO, it's impossible to tell how well Riya will perform in the real world, or how it might compare to other face-recognition technology used by some police departments and government agencies with mixed results.

Shah admits there's room for improvement.

"Riya is but a 2-year-old child in terms of its recognition ability," he said, "but it is ready for you to try it."

Copyright 2006 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.

Discuss:

Discussion comments

,

Most active discussions

  1. votes comments
  2. votes comments
  3. votes comments
  4. votes comments