Entrepreneurs seek new ways to mine Web
Nosa Omoigui, Nervana "... Once we go through the product launch, the demand for the product will skyrocket. " When Nosa Omoigui tells people...
Seattle Times technology reporter
The multibillion-dollar potential of search technology is restoking the entrepreneurial fires of the dot-com boom of the 1990s.
And the Puget Sound region — which played a role in nurturing the early development of search, mainly through the work of students and faculty at the University of Washington — has become one of the hottest centers in the country for the development of search technology.
On the Eastside, in the shadow of a Google development arm and Microsoft's sprawling campus, companies like Nervana are applying search technology to the biotech industry, and InfoSpace is developing it for use on cellphones. In Seattle, Findory is personalizing search results, and Singingfish is searching audio and video files.
And in one of the more intriguing areas of research, a local professor who helped pioneer the field is exploring the notion of combining search with artificial intelligence.
Nosa Omoigui, Nervana
"... Once we go through the product launch, the demand for the product will skyrocket."
When Nosa Omoigui tells people he's from the Seattle area and is part of Nervana, the snickers are inevitable.
He's no grunge musician, but Omoigui's creation is seen as innovative — in the world of medical research.
Omoigui has created a way for researchers to efficiently search through Medline, a massive, government-owned database of documents related to health sciences and medicine. His search engine is so precise, he said, that it can answer such narrow requests as "What is the impact of cell death on lymphatic cancer?" or "Find all research papers on SARS written by Nobel Prize winners."
Search: The engine for change
We have invited search into our lives, and search engine companies have responded by giving us more of what we want — and taking some information in return.
Microsoft's MSN division went from zero to search engine in less than two years — a remarkable achievement. The division did it in typical Microsoft fashion, devoting enormous resources and effort to catch up. The engine still hasn't beaten Google, though. Will it ever?
The Puget Sound has been involved in search development since nearly the beginning. A look at the work being done in search by local people and companies.
In his former job as a researcher at Microsoft, Omoigui worked mostly on television and digital media. Burned out, he quit in 2001 to invent something in one of those areas, but soon he turned his focus to the technology of knowledge.
For information, we go to the Internet. For entertainment, there's always the television and other media. Communication takes place over several platforms.
But what, Omoigui wondered, was the medium for knowledge?
He ran into that problem at Microsoft, where there was no way for one team to know what was going on in the thousands of other teams at the company.
"They had all these silos of knowledge all over the place," he said. "It could be relevant to your job, it could give you new insights, but if you didn't serendipitously stumble upon it, you would never know of its existence."
For months, Omoigui sat in his Issaquah home, listening to jazz and turning the problem over in his mind: How can you search for something you don't know is there? How do you get answers when you have only a vague idea of what you're looking for?
Omoigui, 33, had tapped into one of the most urgent issues in search development. Instead of merely providing links to other Web sites, search engines are working hard at giving answers to questions.
"Whoever is able to provide some lead in this area will be very, very successful," he said.
He picked a niche area — the hot field of biotechnology and medical research — and set out to develop a way to help answer researchers' questions. He named his company Nervana, because he envisioned the service as an information nervous system.
Twelve companies are testing the software, and Omoigui said he'll announce his first customers later this month when his search engine officially launches. About 50 others are planning test runs in the next six weeks, he said.
Nervana has gone from six employees to 13 in the past few months, and Omoigui said he expects the company to be profitable within a year.
"I fully expect that once we go through the product launch, the demand for the product will skyrocket," he said. "I'm very confident about that."
Oren Etzioni, University of Washington
"Imagine a program that is running and constantly learning from the Web.... "
Oren Etzioni could be considered a godfather of search.
The University of Washington professor developed the MetaCrawler search engine in 1995 with Erik Selberg, a graduate student who now works on Microsoft's MSN Search engine.
Etzioni, 41, was trained in artificial intelligence — the idea that machines could develop some level of smarts. His training applied perfectly to search. If computer programs were more intelligent and learned as they went along, they would be better at searching for answers, he postulated.
Now, a decade after MetaCrawler debuted, Etzioni is working on a search engine called KnowItAll that learns as it goes and gives direct answers to users' questions. Instead of listing links to Web sites, KnowItAll aims to read the sites and pull the answers for you.
One day, Etzioni predicts, KnowItAll will give a list of answers to questions like "What are the tallest volcanoes in Europe?" and "What movies does Robert DeNiro star in?" The software is in test form now, and can only give answers to one-word queries. When asked for animals, it returns an extensive list of animal names.
The system is intended to scour the Web, pulling facts and learning along the way so that it can answer a broad set of factual questions. The major search engines can do this now in a limited way and for a pre-set group of questions.
"Imagine a program that is running and constantly learning from the Web and getting smarter over time," he said. "That's very exciting and very difficult to do. Could we build a search engine that learns from the Web?"
The UW has received nearly $1 million in funding for the project from government agencies and Google. Etzioni is getting help from six graduate students, two undergraduates and a research scientist.
He also is developing a computer program called Hamlet that could launch later this year. Hamlet uses software to analyze airfares and predict when airlines will drop their fares.
If Hamlet could find patterns in airfare fluctuations, Etzioni said, it could save ticket buyers money. The technology belongs to the UW, but the university has licensed it to a startup company that Etzioni works at part time.
"My goal is to sort of get the company started and get real business people hired who actually know what they're doing," he said.
Greg Linden, Findory
"People will be ... seeking anything that helps surface what's relevant to them."
Right now, Google's search results for the query mercury are pretty much the same, no matter who does the searching. But Seattle-based Findory rearranges those results based on what a user has searched for in the past. Someone interested in astronomy will see something different than a chemistry student or a new-car buyer.
"This is really the holy grail of search," said Greg Linden, Findory's chief executive. "This is where a lot of people think future improvements are going to be made."
Linden, 32, has plenty of experience with customization, the idea of tailoring a Web site to the user. He spent five years at Amazon.com, a company whose business rests in part on customization, and co-founded the personalization team there. He left in 2002 to get an advanced business degree from Stanford University, and launched Findory in early 2004.
He initially set out to personalize Google's Web search results, but didn't have the user data necessary to do this well. Findory quickly moved into personalizing news articles based on stories the user has clicked on in the past. It searches through blogs, as well.
Findory's only other employee is one of Linden's former Amazon colleagues. The two don't have a formal office, but get a lot of business done at Fuel Coffee, a cafe on Capitol Hill. Linden said he expects to start selling advertising on the site later this year.
The company has spent very little money on marketing itself, but it has tens of thousands of users, and traffic to its pages grows at about 25 percent a month.
A few years from now, Linden said, the emphasis in search will be completely on personalization.
"People will be so overwhelmed with the huge amount of information on their lives they will be seeking anything that helps surface what's relevant to them," he said.
Karen Howe, Singingfish
"All of a sudden
you're starting to look at a demographic that has value."
Seattle-based Singingfish is a search engine for audio and video clips, not links to Web sites, and last week the most popular requests were all for musicians: 50 Cent, Mariah Carey and Green Day.
The site isn't a place to find pirated music, however. Singingfish searches only for legally distributed songs and videos, and as a result many of the clips available are 30-second samples or offerings sanctioned by the artist. The service is a way to at least check out that new song you've been hearing about, or watch a trailer for the upcoming "Star Wars" movie.
Search engines should let people decide if they want to see text, audio or video answers to their queries, said Karen Howe, 49, Singingfish's general manager.
"I see the future as being mixed results between Web-type document stuff and more compelling results that you can get from media," she said. "More companies are going to do this. It's going to be part of their normal collateral."
Singingfish was never designed to be a destination for the consumer. The company, founded in 1999, provided the technology for the numerous directory-style portals that populated the Web during the dot-com boom.
America Online bought Singingfish in 2003 from consumer-electronics company Thomson, and uses the service in the video and audio part of its own search engine, which it has been upgrading.
Singingfish's engine went through a makeover late last year to appeal to Web users, and relaunched with a site directed at consumers. Now users can share queries, refine searches and click an "I'm bored" button to discover new Web sites.
Singingfish charges advertisers a fee to include links to their content in the engine's results, a system called "paid inclusion" that major search engines were uncomfortable using. When users click on those advertising links, they can see messages and options to make purchases.
The site has as many female users as male, and half of them are overseas. Most are online during the day, Howe said.
"All of a sudden you're starting to look at a demographic that has value," she said.
Jim Voelker, InfoSpace
"... The most important device you're going to have in your life is a mobile device."
InfoSpace has a long history of Web search, but lately it's pushing hard to bring that technology to cellphones and other devices.
"If you look 10 years out and maybe even for a shorter period of time, the most important device you're going to have in your life is a mobile device," said Jim Voelker, the company's chief executive.
InfoSpace is betting that people will want to search from those devices as well, but not in the way they do from their home computers. You may not want to know the distance between, say, Mercury and Venus, but you could be interested in finding the nearest gas station or user reviews of that laptop you're eyeing at Fry's Electronics.
InfoSpace went public in the late 1990s, and its stock price soared to staggering heights during the dot-com boom. But it crashed hard amid a cloud of questions about the internal workings of the company's top executives. Earlier this year, The Seattle Times detailed methods the company used to appear more successful than it was.
Voelker, 53, joined the company in 2003, after the previous management had left, and implemented a series of changes that included selling some business divisions to focus on the search and directory business and the mobile-data business.
The company plans to launch a mobile search service later this year designed to get users information in the fewest clicks possible. Wireless network speeds and cellphone power aren't advanced enough yet to make this kind of searching easy, but InfoSpace wants to be ready when the infrastructure gets there.
In the future, that service could be a platform for companies to send targeted ads to users based on where they are at any given time and the things they might want to purchase nearby.
"The nature of the beast will be that you will be very targetable with data you find interesting on the basis of your location," Voelker said.
Kim Peterson: 206-464-2360 or firstname.lastname@example.org