Today marks another milestone for San Francisco based contextual search engine Powerset. They’ve launched a showcase for their user search experience – effectively the search engine minus the web crawl. For now, Powerset queries only Wikipedia and augments results with data from Freebase. The product launch comes just a day after reports that the company is being shopped to potential buyers by investment bank Allen & Co.
I have been able to test Powerset via their labs site for the last few weeks. I wrote about it last month, and the version that just launched is very similar.
There is no way to look at Powerset today and determine if it can be as disruptive to search as Google was when it launched almost a decade ago. That’s because it only queries Wikipedia, and so there is little need for proper ranking algorithms to sort the good from the bad results.
But what user can see is how effective a way it is to gather information quickly. For someone doing research, Powerset effectively removes a number of steps towards getting to the final information. It is particularly effective when the information needed is on many different web pages.
For example, a query on Powerset of “when did earthquakes hit tokyo” yields stunning results. Try this query at Google or even wikipedia to compare – instead of just picking out keywords that are in your query and on a web page, Powerset is actually making some sense of the content included in the wikipedia pages:
The way that Powerset returns queries means that answers are often found in the result snips, as above. They are also structuring a lot of the Wikipedia and (and already structured Freebase) data and inserting it into results. So a search for “Bill Clinton” shows results, but also shows Freebase structured data along with additional query refinements to get to more information. The important thing below isn’t the structured data in the results, its the fact that you can click on the action words and drill down into very specific queries (to find, for example, what bills he signed, or which Supreme Court justices he nominated, or who he slept with).
Powerset is indexing web pages much differently than normal search engines, which generally just record content to match against keyword queries. Instead, Powerset is trying to understand the content on the page so that it can be matched meaningfully to queries later. Even queries that don’t use matching words.
Indexing the web is expensive, though, and Powerset’s way of doing it requires even more time and computing power dedicated to a web page. That’s why they say they aren’t indexing the entire web yet – the company has raised just $12.5 million (plus another $8 million or so in bridge loans from investors). To index the web will require a new round of financing (see the first paragraph above about their sale/financing efforts).
Powerset is has taken a lot of criticism for their goal of trying to redefine how people search the web (including from us). But their lofty goals are what makes Silicon Valley so great – succeed or fail, Powerset is trying to do something pretty spectacular.
The company has also created a demo overview video – see below.
This looks great… can’t wait to try it!
Any idea when a test drive will be forthcoming?
Thanks
It’s live right now: http://www.powerset.com
Looks promising…
My first search was extremely slow though. Maybe just a little techcrunch effect.
hmmm I’m not too impressed. They need to open it up to other sources to become useful…
Nice UI and the results are presented very nicely but I get pretty much identical results if I do a site search through Google. Nice to see that they got rid of that weird logo.
It’s better to come out with a strong showing for specific use, than a weak one for a lot of different uses.
site is very slow and did not impress me. i like a clean look seo this is poorly designed for my taste
just test “where is Paris” and “What is Madonna first name”…not impressed.
Is this is what they have done after raising $12.5 million (or more?)?
I tried this query: “which presidents were assassinated?”. Result is useless compared to Google results.
<>
This actually made me laugh even though I was in a pretty serious mood.
If all it took was more funding Google would have had the solution ready last year.
Powerset doesn’t have a monopoly on genius. No one has an absolute upper edge here.
What it takes is the eventual/gradual move to a more structured Web, but I agree that Powerset is better positioned than Google (less legacy code and better timed entry) to take advantage of the more structured Web that will emerge.
More funding is only going to dilute the owners. I say sell.
Google beater… dream on. Try this simple test:
Who is the prime minister of Canada
http://www.powerset.com/explore/pset?q=who+is+the+prime+minister+of+canada&x=0&y=0
now try on google for a really stunning result:
http://www.google.ca/search?q=who+is+the+prime+minister+of+canada&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a
@10, 11
Try using normal keyword-based query
I have no idea why these guys keep trying to show that they have AI Complete NLP for queries. No one does. No one will for a long time. I’ve never used Powerset, but if you try keyword-based queries then it should work better than a Google search of Wikipedia.
Else, the $12.5M would have been better spent on Falafel
First feature lacking [which I thought should work, or be available in theory]:
Results, which have date information, should be sorted accordingly.
e.g. a query, such as “films starring edward norton by date,” should automatically bring up the freebase results, sorted by date, not sorted arbitrarily.
In my opinion that’s pretty basic. And I am sure I can come up with countless other examples.
I can’t resist a comparison to Dipsie.
Dipsie launches: http://www.siliconbeat.com/entries/2005/11/22/dipsie_finally_launches.html
Dipsie folds: http://www.siliconbeat.com/entries/2006/02/17/dipsie_shuts_already_has_the_luck_gone_at_165_university.html
OK, not entirely fair, since I think Powerset has licensed interesting technology from PARC. But I use the example to make the point that “lofty goals” and “trying to do something pretty spectacular” are a weak substitute for creating value.
Powerset has had access to great money and talent. Maybe something great will come out of it. But so far I’m still waiting.
I tried “Who killed Boba Fett ?” and got some garbage. So much for NLP hype.
This is pathetic. Even “How big is a soccer field?” works better on Google, and that’s exactly the type of query a semantic search engine should excel at.
Michael,
Can you provide some example queries to type in so that we too can have that “AHA moment” !
It’s unanimous – it sucks
This site doesn’t make sense to me. Is the idea here you are going to make sense of my searches when I type in a sentence? How often do I do that? I can’t see how I get anything out of this site I wouldn’t get out of Wikipedia or Google.
Its easy to claim you have the next wave. Not so easy to actually have it. Powerset sure does not have it, as far as I can tell. Then again, Facebook is the #8 site on the net….
this google search:
tokyo earthquake site:wikipedia.org
seems just as good as this crapset search:
when did earthquakes hit tokyo
=== stunning results? what is your infatuation with these guys?
I see the gap between what people expect and what Powerset actually has.
They have something valuable but they’re doing such a shitty job trying to manage expectations. In fact if they had said absolutely nothing and gave no hint whatsoever and just put it out there they would have done far better.
Try using a keyword-based query. Forget NLP on the query side. It’s hopeless and it will be hopeless for a few generations of consumers.
Use a keyword-based query. The results should be consistently better than a “Google search of Wikipedia” .. again: {Better than a Google search of Wikipedia} and NOT better than Google search of the Web.
If my assertion above is wrong then the $12.5M is a total waste.
It didn’t find me in wikipedia; am I so deeply buried that $12M doesn’t surface my remains?
OK sorry, one more:
The question posed to Powerset: what is the best way to make money on the internet?
#1 Answer: Independent music.
Wow. I mean seriously. Even ‘house cleaning’ would be a better answer.
http://www.powerset.com/explore/pset?q=what+is+the+best+way+to+make+money+on+the+internet&submit.x=0&submit.y=0
Interestingly, they seem to be shying away from the one feature I liked in the beta–namely, trying to help users formulate structured queries as they typed them. Sure, their implementation was erratic, but it was the most promising direction I saw them pursue.
In any case, Wikipedia is just a bad set to use for testing question answering (even though I know Powerset insists it is not doing question answering). Almost every top-of-mind question whose answer is available via Wikipedia is answered by a single Wikipedia page devoted to answering that question.
Powerset isn’t ready to advertise on this blog. This PR move shows how desperate they are. $12.5 million, and all they have is this to show for? Even Allen & Co. or Techcrunch can’t save you.
hate these ads. stop it michael
Pathetic, I thought it’s some student summer project. Apply some semantic model on a limited set of quality, semi structured set of of documents in wikipedia is not that hard. 20M down the drain, could’ve been spent better buying food for hungry children in Africa.
actually tried it. 100% off track.
please stop show scripted results. even something like “what is the address of techcrunch” get 100% rubbish. o my god. won’t go back.