Cognition Technologies gets $2.7M to improve semantic search

Semantic technologies have gotten a lot of attention as the potential next leap forward in search, and Microsoft’s acquisition of semantic startup Powerset for more than $100 million didn’t hurt either.

Now a semantic search company called Cognition Technologies has raised an additional $2.7 million in funding. Chief executive Scott Jarus is happy to tout Cognition’s technology (more on that in a second) and to praise the competition (which also includes Hakia, Expert System and others), but interestingly, he also argues that if semantic technology is going to make a real “quantum leap,” it’s going to require a bigger roll-up eventually by Microsoft, Google or another major tech company.

“We’re all heading in the same direction, but we’re driving different cars, none of which will be able to make it to the end by themselves,” Jarus says.

The Culver City, Calif. startup is approaching the space a bit differently than Powerset. For one thing, it isn’t trying to create a search engine per se, but rather sell its semantic technology to other companies. You can already see Cognition Technologies in action at Lexis-Nexis Concordance (a litigation database), Medline (the database of the National Institutes of Health) and elsewhere.

The technology is different, too, Jarus says. Where Powerset focuses on the structure of a search query, Cognition Technologies has been building a “semantic map” — basically, a big dictionary that allows Cognition Technologies’ products to actually understand the meaning of the words in your query, and therefore understand what you’re actually asking for. The startup has even published a white paper comparing its technology to Powerset’s. In the paper’s first example (and others), a Wikipedia search for “Who won Wimbledon in 1956?” returned more relevant results through Cognition. GigaOM also compared the two companies recently in “a semantic search shoot-out” and found that Cognition does better when “hardcore semantic parsing” is required — in other words, when the query is particularly complicated.

To improve its offerings, Cognition is developing a new semantic parser, which should beef up its ability to understand the structure of different queries. Cognition Technologies has been around for more than 20 years, but only took its current form about five years ago. Since then it has raised a total of around $6 million. The current round includes Draper Associates, Fingerhut Ventures and a personal investment from Jarus himself.

Next Story:
Previous Story:

Tags: , , ,

Photo of Anthony Ha

About the Author, Anthony Ha

Anthony is VentureBeat's assistant editor, as well as its reporter on enterprise technology, cloud computing, and tech policy. Before joining VentureBeat in 2008, Anthony worked at the Hollister Free Lance, where he won awards from the California Newspaper Publishers Association for breaking news coverage and writing. He attended Stanford University and now lives in San Francisco. Reach him at anthony@venturebeat.com. You can also follow Anthony on Twitter.

  • jamalystic
    I believe that search is in its infancy and that there is still much grounds to be gain. So it's nice that search companies age getting the financial boost they required in order to further thie rdevelopment:Searching for the Next Search Engine(http://www.internetevolution.com/author.asp?section_id=466&doc_id=152415&F_src=flftwo)
  • millsdavis
    We compared Google, Powerset, and ExpertSystem. Poweset's UI is better than Cognition (at this stage). Expert System's performance was better in our tests than either.
  • Hmm, can you point us to more info about your tests? I'm particularly curious about your comment about UI; I'd think that's a less important issue for Cognition, since it's not building a consumer-facing search engine, but providing the underlying technology for other products. Then again, I'm not a terribly technical guy, so maybe I just said something really dumb.
  • millsdavis
    Anthony,

    Our test was pretty limited. We posed only a handful of queries
    selected so that some were more factual, some required more "semantic"
    interpretation. For each query we measured input time, review time,
    accuracy, and misinformation or bias.

    You're probably right. Cognition has not emphasized UI considerations
    as much as Powerset. On the other hand, I would always argue that UI
    is always on of four key dimensions of product value -- unless you are
    selling a service with an API.

    Mills
  • scottjarus
    Mills,

    We'd be happy to set up a briefing for you regarding Cognition's business model, mission, and technology.

    When it comes to a comparison between Powerset and Cognition, we typically direct readers to a Cognition vs. Powerset comparison white paper that we were asked to write shortly after Powerset's launch:

    http://www.cognition.com/Why_Powerset_Misses_th....

    But suffice it to say, Anthony is correct. We are not focused on being a "Google Killer" (unlike Powerset). We are a semantic enabling technology which enables other applications and technologies understand the meaning of language.

    Through our patented Semantic Natural Language Processing (NLP) technology, which was developed over the past 23 years (300+ person years), Cognition delivers significantly higher levels of relevant text understanding than is possible with currently used NLP and/or Search technologies. Unlike other Semantic NLP companies, Cognition is the first and only company to combine all the right linguistic elements to optimize semantic understanding, including advanced parsing technology and the most comprehensive Semantic Map of the English language available today.

    --Scott Jarus, CEO, Cognition Technologies
  • edhardy622
    My girlfriend bought me a pair of Chestnut color UGG boots short for Christmas.
    http://www.uggboots365.co.uk
blog comments powered by Disqus