RealTime IT News

'Future of Search Will Make you Dizzy'

NEW YORK -- Amazon.com's A9 subsidiary wants to play a large part in pushing search technology to a future where the relevancy of search results will be startling and exciting.

That's the word from A9 chief executive Udi Manber, PhD, who insists that full development of search and resource discovery tools remains at least a decade away.

"Think about how the Web has changed your life in the last 10 years. Now, try to extrapolate 10 years forward and you should feel dizzy. We're still in day one of developing and innovating in search. There's still a lot of exciting discoveries to be made," Manber said in a keynote address at this year's World Wide Web (W3C) conference here.

Manber, who worked as Amazon.com's chief algorithms officer before taking the reins at A9, predicts a future where the relevancy of search results will be measured and understood to deliver information to users.

"Search is a huge area and we have made a lot of progress but there are still a lot of things to be done. Despite all the advancements, the truth is that we still can't find what we're looking for," he said, making it clear that his company was not trying to duplicate the work of Google.

"A9's mandate is to build new search technologies to improve the user experience. We want to invent new things and new ways of finding relevant information. The first question I get from people is, 'Are you going to build another Google?' But, no, that's not what we are doing. There's so much room for innovation that you can build interesting things that aren't available today."

He said he believes that user-dependence on single-word search queries present a "huge barrier to advanced technologies" and called on developers and researchers to avoid the trap of giving up relevancy at the altar of increased speed.

"For most users, they expect it to be as simple as possible and that's a barrier. If music was invented 20 years ago, we'd all be playing one-string instruments," he said, suggesting that user habits needed to change to adapt to the advancement in search technologies.

Another hiccup for researchers, Manber said, is that the relevancy of search results is hard to measure. "Relevancy changes all the time and is not well understood. Relevancy is different from user to user. We have to figure out better ways to measure [results] to make it better. That's the hard part. We need a science around measuring relevancy."

"It's not about speed or size anymore. It's all about quality. It's about delivering the tools that allow relevancy. It's good to make searching faster and faster because that part is well understood. The quality part is not understood and that's the challenge we face today," he added.

Manber, who is credited with creating Amazon.com's widely hailed "Search Inside the Book" tool, said A9's developers had the luxury of long-term planning, which is crucial to the R&D process. "I don't have VCs on my back asking about revenues. I can think long term and spend time experimenting to build production quality services."

Manber described "Search Inside the Book" as the most exciting project he had ever worked on and gave an overview of process of scanning 33 million pages from more than 120,000 books while creating the text-search tool. "There were lots of monitoring tools, lots of testing, lots of databases and lots of support for huge data. The entire thing took about six months to launch and it's something I'm very proud of."

The search technology pioneer also outlined the personalization capabilities of the A9 portal, which is integrated into Google and Amazon.com to allow Web and book searches in real time. It is also programmed to save user searches into personalized profiles.

Manber is no newcomer to the search space. He was a lead developer in technologies such as the Search Broker, which provides a two-level Web search paradigm by forwarding each query to a specific search engine and Web Glimpse, a search tool that provides a flexible combination of browsing and searching.

A former professor of computer science at the University of Arizona, Manber has done extensive development work in search and resource discovery tools, software tools, computer networks, computer security, and design of algorithms.