Semantic start-up takes aim at Google search
A SILICON VALLEY start-up has just unveiled a new search system which in time should make it possible for people to search the entire Internet using natural language phrases and questions, instead of restricting them to keywords. Powerset has launched its new tools on the online encyclopedia, seen by some as a microcosm of the Internet, Wickipedia.
What Powerset’s tools do, is they take words and sentences entered by a person in the search field and look for related concepts to those words, thereby expanding the search scope extensively, making search more flexible, and answers easier to find. In other words, it doesn’t tie you down to your keywords, it can figure out what you mean.
With its new flexible word in context and concept search, it looks like the small 60 worker, two and a half year old San Francisco based start-up company could be well on its way to becoming a major challenge for the likes of Yahoo, the Vole, Ask.com and even search engine giant Google.
An Internet analyst for Sterling Market Intelligence, Greg Sterling, told Reuters that "This could become the basis of a Google-killer". Sterling added that the big threat lay in the possibility that a company like Microsoft, which currently has a spare $44 billion burning a whole in its pocket after the Yahoo deal collapsed, could snap up Powerset, hence becoming a very formidable rival indeed.
But for now, the company says it is content with using Wackypedia as a “look what we can do, aren’t we clever?!” showcase, with aims to expand its operations to the CIA world Factbook, financial or patent filings and wackpedia-likes in the near future.
The company has also said that it will attempt to join forces with several other high-quality data sites where data can be organised in question and answer form or that is compatible with Powerset search techniques.
At the moment, Powerset is particularly focused on getting its technology perfected as well as properly up and running. This means that almost all its staff are either computer boffins or linguists, leaving little room for marketing execs or ad pushers, but this will not always be the case, sadly. The company has said that it eventually plans to sell advertising space which would display next to search results. Google, watch your back. µ
L’Inqs
Reuters
Share this:
ShareBusiness Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
Go on, ask it "How long do ferrets live?" the first two answers involve Sylvester McCoy and Rudi Guliani. Only the fifth result mentions ferrets in context and even then fails to answer the question. Complete twaddle!
Type in "Is Microsoft evil". Note the second link. What a peculiar Vista.
@Geoffs:

They´re a STARTUP company. And, considering what they´re attempting, I´m not the least bit surprised if it starts out as inaccurate. If they can supply an engine that can answer 1 out of 2 questions accurately, they´ll have a major accomplishment, in my opinion.

And if they can make it even partially accurate, then it´s just a matter of time before they make it super-accurate.

I can imagine a situation where they have a free version that only researches your question for anywhere from 10 seconds to a minute, and a paid version that asks for a question and an amount, and attempts to answer the question for that amount. (Actually, it would probably be much more complicated, but I don´t want to annoy people with my analysis)
Thought for a horrid moment that headline read "Symantec start up takes aim at Google". 

The idea of a Norton search engine taking over my computer....
@Jason - Either its semantic search or not. The question is simple but quite specific because "ferret" is a rare word and "how long" is a specific type of question, if this thing is working in any state it should be able to discern the answer from the input. If they're working on it and its in this piss-poor state then they certainly shouldn't be publicising it and trying to wrangle $m out of investors.
--[And if they can make it even partially accurate, then it´s just a matter of time before they make it super-accurate.]--

That is not true of every problem, particularly not the sort of problems AI turns up. See also: weather prediction.

Having said that, I typed in "ducks" and it said "We know factz[sic] about duck! Uses: bombs"