Real life is in. Just as audiences are eating up the latest amateur-captured video of government and civilian misdeeds or wire-tapped cellular phone miscues, mere mortals are fast being mined for "content."
Nowhere is this idea more evident than in emerging services such as Fab, an experimental system that selects Web pages for a community of 100 Stanford University users based on their own preferences, as well as on the profiles of other members with similar interests.
"It's weird that you have so many individual users on this worldwide network," said Marko Balabanovic, a Stanford graduate student who showcased Fab for the first time on Friday at the First International Conference on Autonomous Agents in Los Angeles.
"[With Fab] there is a potential for generating a relationship with other people who have similar interests."
Netizens want powerful artificial agent tools such as Firefly to snatch up the wheat from the mass of chaff that is the Web. But the problem with traditional information retrieval systems such as the Mercury Center's NewsHound is that they are static. Balabanovic believes the agents in these systems must evolve to reflect people's changing tastes.
For example, a member who is interested in music may receive pages on ambient music or jazz, or Hole or Whitney Houston. With each wave of selections, users give the agent feedback on specific pages. So a user might pan the Whitney Houston page and approve the jazz selections - teaching the agent that the member may not like pop music but loves jazz.
Fab also creates community-based agents that are not specific to any single user. They search for Web pages based on the collective preferences of a given community, making assumptions on what a user might like based on interests it determines are common with other users.
But to take advantage of Fab's agents, users need to spend time training them. The same is true of Firefly. "It's a lot of effort," Balabonovic concedes.
Enter Imana. The year-old San Francisco-based start-up - which has hired Balabonovic - will launch an automated version of a service called SiteSeer in the second quarter of this year.
SiteSeer extrapolates individual and collective interests based on information it culls from bookmark files volunteered by its members. Where Fab identifies interests and delivers the most relevant Web pages, SiteSeer plays off the interests expressed by users as they select pages.
"We have a gazillion people banging on the Web," said Imana president James Rucker. "It's not just me doing a search and not just me collecting bookmarks. There's an opportunity to help each other out and share discoveries."
With the launch of SiteSeer, Rucker and Imana chairman Marcos Polanko foresee the start of a series of SiteSeer communities, each organized around a shared interest.
But such community and companionship may come at a cost. The SiteSeer prototype requires users to voluntarily upload their personal bookmark files.
The upcoming version will make this a passive process, without compromising the privacy of users, assured Polanko. The new SiteSeer, which will work on the Internet and intranets, will reportedly prompt the user for approval before publishing a bookmark file on the server.
