Do a Google search of that company”™s own name and you get 1.4 billion hits. Do the same on Semantifi Inc.”™s own search engine, and you get just five.
Give Semantifi CEO Shree Pragada 14 months, and he promises you will find things on Semantifi.com you never would on Google ”“ assuming the Stamford-based startup stays a step ahead of the Silicon Valley giant.
Semantifi”™s innovative approach to search technology is already drawing eyes, with the company the lone Connecticut startup to participate in last month”™s DEMOfall 2010 conference in Santa Clara, Calif., among the most closely watched high-tech confabs in the world.
The company”™s portal is designed to crack open access to what search engine companies dub the “deep Web,” which accounts for 99 percent of the information on the Internet but which is largely passed over by Google and Bing”™s Web crawlers.
Adapting the Wikipedia model
The term “deep Web” is attributed to Michael Bergman, who published a 2001 paper in the Journal of Electronic Publishing that analyzed the problem. Nearly a decade later, Semantifi thinks it has found an answer in the Wikipedia model, in which users contribute their own input.
In the same way anyone can create an encyclopedia article on Wikipedia, anyone can link an accessible database to Semantifi. Hosted by government agencies, businesses, or individuals, many of those datasets escape the attention of traditional search engines, according to Pragada and other deep Web devotees.
In theory, the Wikipedia model should allow Semantifi to extend its reach faster than a small team of Web programmers could do on their own as they attempt to keep up with a constant stream of new sources of data.
The question becomes how effective the company can be. A late August search of the word “Semantifi” on Google dredged up 11,800 returns; when the company”™s name was entered into its own website, it returned a message reading “Unable to recognize query. Please reenter.”
Attracting posters, content quickly is key
Pragada acknowledges the importance of the company adding a large number of datasets in a short period of time, to give early users success using the website even as the company and outsiders add more. He adds that the quantity of search returns served up by Semantifi is less important than the relevance of those returns, which the company is attempting to improve through software innovations.
For instance, searches on Bing and Yahoo using the term “Semantifi” dredged up 8 million results, though cluttered with similar sounding names and words, such as “semantic” and “Symantec Corp.” In much the same way, Pragada noted, searches on standard search engines using more-complex phrases often miss the mark, forcing users to scroll through page after page, or to refine their search terms in differing ways ”“ sometimes with success, sometimes not.
Myriad companies have worked on the problem of applying advanced semantic capabilities to database searches, but few have attempted to build a Web search company from that effort. Google itself reportedly has internal teams attempting to improve the company”™s access to the deep Web.
Founded as ExeCue in 2005, since then Semantifi has raised nearly $5 million in financing, with its backers including the Connecticut Innovations Inc. venture capital fund backed by the state.
Pragada attributes his company”™s presence in Stamford to Connecticut Innovations”™ early support ”“ how fast it will grow is completely dependent on how quickly it can get the broader universe of Internet users linking in data.
Wikipedia, at least, has provided the company with a roadmap to how that can be done.
“Here, we are at least learning from the other revolutions,” Pragada said.
