Searching for something on the Web can be frustrating. Too often, search engines turn up no hits, or they come back with page after page of unrelated junk.
"Search stinks," is how Jeffrey M. Stibel puts it. He is the 26-year-old founder and chief executive officer of Simpli.com, a small, Providence-based Internet startup.
Stibel's company, now about a year old, is trying to make searching the Internet more fruitful. The company has developed technology it hopes to license to search engines and Web portal sites.
What's interesting about the firm's work is that it has used old-school academic thinking to develop cutting-edge technology. Simpli.com applies some long-established principles of cognitive science and linguistics to help improve Internet searches.
What could cognitive science, the study of how the brain works, and linguistics, the study of human language, have to do with finding something on the Web?
Plenty, says Stibel, who earned a master's degree in cognitive and linguistic sciences from Brown University last year.
The fundamental problem with search engines, he says, is not that they are bad. It's that people and computers communicate in completely different ways.
People, often without realizing it, rely on context and the environment to help get their message across to others.
For example, in a coffee house, if someone says the word "java," it's very likely they are speaking of coffee. But in an office setting, at a meeting of programmers, someone who mentions Java is more likely talking about a programming language. People get the implied meaning.
By contrast, computers don't just get it. They don't have the luxury of knowing context. Entering "java" in a search engine will likely bring up hits about coffee and programming, and a host of other meanings, including Java, the Indonesian island.
That's where www.Simpli.com
"We're trying to bridge the communication gap between people and computers," Stibel said. "We believe that is the fundamental problem."
At the Simpli.com site, the company has set up its own search page, which employs its technology, called SimpliFind. The site taps into several different popular search engines to display results.
Here's how it works. After entering a search term on the Simpli.com site, it tries to clarify what you are looking for by presenting categories your search may fall under.
For example, if you enter the search term "Mars," you are given a selection of possible meanings. You can select those SimpliFind has come up with -- Mars the planet, Mars the candy company, Mars the Roman god, and other meanings -- or enter your own.
SimpliFind makes these associations using a massive "knowledge base," a collection of words and how they relate to each other. Select Roman god, and click search, and that's where the second half of Simpli's technology is put to work. SimpliFind assembles a "query string" that gives search engines detailed instructions to help narrow down the search. In English, the string might translate to something like "find Web pages with Mars the Roman god and exclude references to Mars the planet." The actual string uses the arcane syntax that search engines understand, and contain number "weights" used to emphasize certain terms and de-emphasize others.
Then, behind the scenes, SimpliFind sends its specially constructed queries to several different search engines. It then displays the results neatly, as if they came from a single source.
The limiting techniques are not foolproof. If you select Mars, as in the French word for the month of March, the third hit Simpli.com returned last week was for NASA's Mars Polar Lander Web site.
Nevertheless, Stibel has marshaled a team of believers. He persuaded James A. Anderson, head of the Department of Cognitive and Linguistic Sciences at Brown, and some other top scientists and programmers at Brown, to join the company.
Anderson, who at 59 is more than twice Stibel's age, took a leave from Brown last semester to work fulltime at the effort. He is a vice president and chief scientist.
"This is the most exciting thing I've done along these lines," Anderson said. "It's been great fun."
One of the company's scientific advisers is George A. Miller, a researcher in psychology and linquistics who headed a 15-year old project at Princeton University to build WordNet, a database of words and relationships between those words. Simpli.com based its knowledge base on WordNet.
Stibel said the company is building its staff quickly. There are about 22 full-time employees now, but by next month, Simpli.com plans to hire 8 to 13 more. The company plans to move from its cramped quarters in the basement of a building on South Main Street to a space seven times the size at Davol Square. He says he hopes the move will be finished by July 1.
The company has raised about $500,000 in seed money, primarily from employees, according to Peter Delgrosso, director of marketing. Simpli.com is negotiating with other investors, he said.
Stibel said one hurdle the company faced early on was, of all things, sleeping accomodations for employees.
"At the end of the day, people don't really want to go home," he said. Some employees began sleeping under the conference room table, Stibel said, which was a signal he had to do something.
One evening, on the way back from a meeting with a venture capital firm in Boston, Stibel stopped at K-Mart in Attleboro and bought two sleeping bags.
He was greeted with applause when he brought them back to the office about 11 p.m. that night.
Timothy C. Barmann covers technology for The Journal. His column runs every other week on the More For Your Money page. He can be reached at email@example.com.