correlate Rotating Header Image

Natural Language Search & Powerset

Jun 22nd, 2007
by Lou Paglia.

PowersetThere has been a lot of pre-coverage of Powerset, starting with the TechCrunch post back in February. Many claim that they have something in development, to be released for public view in September that can change the tide of the search marketplace. And they will do it with natural language search.

I have to say that I will believe it when I see it, that is my stance right now. Not that I don’t believe a company can develop a category leading application or service even when the market “seems” closed out. I made that mistake once before in 1998 when someone first introduced me to a company called G-O-O-G-L-E, and Yahoo! seemed like that had the search market monopolized.

My curiosity around Powerset is more around the paradigm of natural language search than the company itself. Any time you ask people to switch their behaviors and usage habits, you are setting up another barrier for yourself. People are used to typing a few words or even advance queries into the “little white box”. Will they shift to typing in “Who won an academy award in 2001?” like discussed in VentureBeat earlier today. Time will tell but right now most would type “academy award 2001″. It will be critical in how they overcome that challenge.

I, for one, am very interested to get a peak into what they have developing and signed up for Powerlabs today. Bringing the early adopters and users into the process of testing out various elements in a fantastic way to get early feedback, hone the engine on obvious user tasks, gain early market momentum and build word-of-mouth. Smart move (if the engine provides a satisfactory experience and doesn’t not turn users away), users rarely come back if the first experience is a bad one. And in the case of Powerset, the experience is going to have to be better than Google on the first pass, otherwise users will go with what they know.

To close, let me use an example.  On the Powerset blog,  they use the example query “Who proved Fermat’s last theorem? in a post.  Their results set are impressive providing the instant answer.  Just for validation, I searched “Fermat’s last theorem” at Google and a Wikipedia page came up as the first hit.  The answer “Andrew Wiles” was found in the sixth sentence.  Great example of two things.  First, showing Powerset as an answer finder, and second, raises the question if search engine switching costs are low enough that a little less legwork to find the answer is going to win the hearts of users.

Sphere: Related Content

Posted in: Uncategorized.
Tagged: · ·

← Good Article on Steve Jobs
Facebook and AOL →

Viewing 2 Comments

    • ^
    • v

    Lou, I think you are right to address the question of switching costs, which are related to what the workplace anthropologists call "immutable work practices." Remember the days of the jokes about expert systems being solutions in search of problems? There is no question that there are problems that can only be solved by "deep reading." My guess is that you could rattle off a few of those problems from your own experiences. I might also guess that, for at least some of those problems, you could tell stories about the work practices that emerged for dealing with them and how effective those practices turned out to be. So the real question will not have to do with finding answers but with whether the linguistic technology behind Powerset is (significantly?) more effective than practices that are already in place.

    • ^
    • v

    Yes, I think that is definitely sound thinking in the workplace. Many of Factiva's most powerful users write extremely advanced queries to get to the information they need or the information that is most relevant. Many queries could be considered small applets within themselves.


    The consumer space, for me, is the bigger question. Is the linguistic technology going to be better enough for people to change their search habits. Consumers (majority of them) aren't interested in changing their ways from something that "basically works" if it isn't easy. I'm an early adopter but I really have no interest in typing questions (and remembering to do so). But as an early adopter, I will certainly try it when it is available.

 

Trackbacks

(Trackback URL)

close Reblog this comment
Powered by Disqus · Learn more
blog comments powered by Disqus