September 17, 2007

Powerset launches Powerset Labs at TechCrunch40

SF BetaPowerset is undertaking a huge task: building a natural language search engine that reads and understands every sentence on the Web. The good news is that thanks to the technology that we’ve licensed from PARC combined with homegrown technology developed by our strong team of linguists and computer scientists, we’re well on the path to achieving this goal.

We realize that most companies wait to launch until they have a completely usable beta version. Because Powerset is a natural language search engine, the earlier we have input from the best natural language processing units on the planet – the brains of humans – the quicker our search engine will improve. Through a combination of quantitative feedback, qualitative suggestions and AI learning techniques, Powerset will get much smarter when people are interacting with it.

That’s the reason we’ve decided to create Powerset Labs, which we are opening up to the first group of users today at TechCrunch40. Powerset Labs is a community where users can provide feedback on our product design and natural language engine. While users will not be able to interact with the Powerset open search box across the Web, we are giving users a peek at technology demonstrations that show off some of Powerset’s natural language processing capabilities.

Though the content of Powerset Labs will change based on user feedback, we wanted to share with you what we demoed today at TC40: Powermouse and Use Cases.

Powermouse is a window into Powerset’s natural language index. When Powerset reads sentences in Wikipedia, we go from open text to representations of meaning. In other words, we take text and turn it into structured “facts.” When users enter a query into Powermouse, they’ll be able to browse the “facts” stored in our index. In the example below, when wrestling star “Hulk Hogan” is entered in the first “something” box, users can see all of the facts we’ve indexed about him. Now, if you add “defeat” into the connection box, users see all of the facts that we’ve indexed from Wikipedia about wrestlers that Hogan has defeated. Here’s a Hulk Hogan screenshot. In addition to showing off the power of our index, Powermouse also shows a different type of interface that’s possible with a natural language index.

Uses Cases demonstrates how a natural language query can exploit Powerset’s index. Unlike Powermouse, Use Cases lets users express their intent in natural language. We’ve picked about a dozen use cases that illustrate how a natural language index can return results that are qualitatively different than keyword results. For example, here’s a query of “Who mocked Blair?” that shows how Powerset understands all of the various ways “mock” can be expressed in English. After a query, we encourage users to tell Powerset which of our results are good and which are not. We also ask that users vote on which results are better: the Powerset results or the keyword results on the right. All of this user feedback is what will help make Powerset a better search engine.

Once users have tried out the applications in Powerset Labs, we invite them to submit ideas about how to make them better. Within Powerset Labs, users can browse through ideas, vote on the best ideas and comment on ideas. As users participate in Powerlabs, they’ll get karma points for everything they do. Eventually, users with the most karma will get perks within the community. The bottom line: We’re listening and we’ll try to implement your brilliance as soon as we possibly can.

As a note, Powerset has received a lot of attention over the past few months and we’ve been overwhelmed with the number of people who have signed up for Powerset Labs. Instead of letting everyone in at once, we’ve decided to let people into Poweset Labs in the order they signed up. We want to make sure that each group of Powerset Labs users gets a great experience, so we’re going to grow the community slowly and carefully. If you’d like to sign up, go to labs.powerset.com and we’ll be letting in the next wave of users as soon as possible.

We’re really excited to see the Powerset Labs community grow, to gather and implement your feedback, to share with you more and more technology demonstrations and prototypes, and ultimately to deliver a transformative search engine to the world.

Oh, and we have our first official social media press release about our launch at TechCrunch40 if you are press and need contact information, quotes or screenshots.

Comments (119)

So when do we get to see this amazing feat ?

Can’t wait till this comes out :-)

Hopefully this’ll be better and safer than Google ^_^

You may not even have estimated the appeal of a Google alternative for those of us disenchanted with Google’s corporate social and cultural stances and activities. I believe there are millions of people who want a search without the ‘tude.

WOw this sounds sooo awesome!! I cant wait until it comes out. nods

I just did a Google search for “birds” and the result indicated there were about 116,000,000 possible sites. Why doesn’t a search engine just list, say, the top 20 or so hits (about 2 pages) and keep track of the ones that are then viewed the most and put them at the top? The poor performers could be dropped over time and new sites added. I think 99% of users would find what they need or be more or less happy with what is found in the top 20 results. For the 1% who don’t find their answer, well, too bad. Search engines try too hard to be all things to all people.

@ Mary: sometimes, the hardest problem people have with search engines is trying to locate that hard to find 1% area that just doesn’t match their keywords. Unfortunately, search engines don’t operate on the same principles as Top 40 radio.

I’m curious though, what are the plans for non-English queries?

whoops, i meant @ Omar… darned confusing signatures >.>

Using the response from the beauty contestant from South Carolina as an example of parsing language is creative. It would have had more impact if you had spelled the state correctly within the body of the paragraph. On my screen, it appears spelled correctly in the deading, but within the body is spelled “Carololina’s”.

Devil in the details kinda stuff, I admit. But, it did jump out.

My apologies. delete the word “deading” in the last sentence, and insert “heading”. I couldn’t find the edit function.

No hits should be removed. What is needed is the ability to jump around within the list. Such as going to the end of the list to see what is deemed of lesser importance. It is also sometimes desirable to do a search within the list.

I can’t wait for this to come out. Why didn’t anyone think of a natural english language search engine before? Maybe if this becomes really popular, they will add more languages (probably french and spanish first.)

This is not by any means a new idea. Ask Jeeves was the first major “natural language” search engine, but it was horrible and failed miserably to produce relevant results. I highly doubt that this will overcome Google in the long run. Google has invested over $2.2 billion in R&D for search technologies since 2005. I’m sure that somewhere in that huge pile of cash is money for developing natural language queries. I bet it takes no more than a year for “Ask Google” Beta to launch in Google Labs.

Good.. seems to be the next logical step in search.. add semantics and intelligence… Worked on a concept called Case Based Reasoning long long time ago.. looks like those pricicples are used in this..

Looks very professional! I am a big fan of Clusty, which was developed at CMU. Given the refinement I am seeing, this looks even better.

What is Powerset’s policy on collecting data from your users and then giving or selling it to government agencies?

I am looking for an honest answer not one that has been sterilized.

Peace Carlos

the real web2.0 is about to born.. I hope :o)

I don’t get this. When I look at my search history in google, hardly anything I’m searching for would be better expressed as a natural language query. Are you expecting that powerset is the one-off thing that gets used in those rare cases where natural language is appropriate? I don’t think there’s anything bad about keyword searching when I’m trying to find a professor’s website (just enter their name) or a page about the end of old-time country singer jimmie rodgers’ life (another gem from the last few days). In fact, I think keyword search is terser and more efficient for me. Why would I want to write “what is barney pell’s homepage” when I could just write “barney pell” on google (it’s the first hit, by the way).

I wish you guys luck, but I think you should expect to do pass-through to google for queries that aren’t looking for a specific piece of data. Speaking of which, google does seem to handle that, too… When I enter “when did jimmie rodgers die?”, it comes back with

———————- Jimmie Rodgers (II) — Date of Death: 26 May, 1933

According to http://www.imdb.com/name/nm1034229/bio - More sources » ———————-

Call me a skeptic, but I’m skeptical…

As a scientist I hope someone comes out with a way to search within defined parameters. In google even if you define a subject and try to search within the subject, most of the crap comes along. Good luck. Keep us informed. Glad to beta test if you have something.

Perhaps an interesting blog post: how would Powerset parse this sentence? Would it work? http://en.wikipedia.org/wiki/Buffalobuffalobuffalo

Thanks.

The time I would save will be incredible. If this works I’ll be there. Please keep the commercial content to a minimum.

AS A RESOURCE SPECIALIST I’M GOING TO LET ALL OUR CLIENT USE THE SITE FOR ANYTHING POSSIBLE. I CAN’T WAIT TO START USING IT I’M SURE IT’S GOING TO BE THE BEST.

Cuando podremos usarlo?

I have been looking for a search engine that would give answers to specific question instead of thousands of sites vaguely related to my questions that most other search engines go at present.

I wish you all the best for a successful launching of this useful search engine.

Don’t create interest you can’t fulfill.

You’ve created buzz. We came. There’s nothing.

Probably wont be back, too bad.

Do you have any plans to make a Public offering of your stock, like Google did?

This will be wonderful for those times when you don’t exactly know what you are looking for and keywords don’t work. Hopefully you will be able to type in a fairly long description of the item and get an answer. I often have to look for things that I have no idea what they are called, and sometimes by accident find the proper name after searching fruitlessly for hours. Let’s give it a good old try and make it work!

Yo Powerset;

I hope it works out. I am a little tired of going through thousands & thousands (not really) of pages to find an answer. But you get the hint. I am NOT a ‘Whiz-kid”. Please keep it simple

Thank you Jerry

This is an exciting project. Yet I hope every web page will get get ‘fair’ result as they are searched.

N

This proyect is the simpliest thing on the web,, anyway, i guess that when you put the question on the searchbox your search engine will take the keywords and search the web,,

its also possible to put questions on google,, whatever you put, they will make almost the impossible to give you what you are searching for,,

there is no one like google.

As technology provides a means for increased available information, and at a blink of an eye….our patience to attain the information apparently decreases. Keep it simple to use, with quick results….and it will be used! Larry

Hi, Will it give powerset also into German language?

kind regards Tom

If you want to improve on Google, make the results of the search sortable. I would give anything to be able to sort results by date.

Simple, to the point and we don’t need a million results!

Way to go, I’m waitin’ for this to come out.

-AB

nature language search and communication, we have heard for many years, and seen in the movies (terminator, AI, matrix..etc) ..

I, also tried the ALICE project, and have built a chinese chatbot for myself …

but the turth is, I am not happy with the result .

Hope poeple in Powerset cna give me a surprise this time!

Good luck !!

toi thich duoc di hoc va ngoi truoc chiec may vi tinh

Serving only one language is your weakest point and making you incomparable with google. Start your research on other languages like spanish and chinese asap, or you will just be one of the AI engine supporting one particular language in the world. let me know if u need support in chinese.

New Age life’s work Book of Changes tarot sense light 塔羅易經光的交會,林老師態勢自覺諮商、創作和教學。

New Age life’s work Book of Changes tarot sense light 塔羅易經光的交會,林老師態勢自覺諮商、創作和教學。

look forward to use program

:?)

Eagerly looking forward to get my hands on this new technology

i think i can help with spanish sentences..

i can smell a great project :)

I realy have no any idea what will be happened in the near future. Because I can feel this new search engine will be change the world soon.

Hahahaha……

Francis

Wow! who would have thought that the day would come when someone else is on the verge of taking down the big “G” and other SE within the next few years.

It just goes to show you that this is the way natural search will evolve.

Great job guys, to your continued online success

Wow, I expect to use your amazing technology for searching through internet. I think this technology is soooo powerful and save much time to scan what I would want to seach from the search result.

Cheers!!!

I could’nt wait. I am excited by your new search engine. Let the word out. You could send your link to my website www.lanewsweek.net and www.azurelreport.com I will put your company for 1 year for FREE.

Sam

I can’t wait to see this baby released. Business Week just said that Powerset should be up there with Google and possibly even better.

I suppose that when you search based on meaning rather than context you will most likely get better results.

Bravo!

Buena iniciativa, os felicito porque vais a tener El Mundo en Tus Manos

The results will be the tell so to speak. The complexity of search can be tested by a simple iteration through an expanding set of nouns and verbs added to the base search criteria. Mary above was searching for birds which is a very big topic - we don’t want to lose references at all - we need to narrow the search and do it very intelligently.

We don’t ask the library to throw out the 3 shelves of books on birds we don’t need we just narrow our search until we find the flightless birds of Australia.

hits on Google by way of example birds = 110,000,000 birds that cannot fly = 2,140,000 birds cannot fly = 2,070,000 birds cannot fly Australian = 1,890,000 birds cannot fly Australian carnivorous = 238,000 birds cannot fly Australian small carnivorous = 235,000

so we can continue to narrow our search but the result is always based on the same algorithm - look for occurrences of documents that contain all of the words, some of the words etc…

The search engine does not take into consideration Mary’s state of mind i.e. her frustration at still having 235,000 results after having refined her query 10 times.

IMHO the search engine should understand and anticipate the users state of mind based on the progressive multiple attempts at search refinement, it should search for context and word association including lexographic proximity not in the document being searched but in the search term being submitted, the engine should also consider the order of request expansion to follow the thought process of the person searching. If Mary starts out with birds and progresses to Australia last it means there is more interest in birds than Australia and likewise as flightless was entered before Australia there is more interest in all flightless birds than there is in Australian flightless birds. Additionally synonyms and word variants should be automatically and maybe invisible added to the search criteria to assist the user in refinement.

birds cannot fly Australian carnivorous flightless = 15,700 hits

The human brain can easily translate the idea of “cannot fly” into flightless but unless I specifically include it in my query the engine doesn’t add it even though it has the same meaning.

So to end my rambling diatribe It will be interesting to see this engine in operation, is it another natural language reader or is it actually going to exhibit some intelligence ?

Looking forward to testing Powerset. It will be interesting to see this engine in operation. I will be interested to see what “intelligence” it might show in answering my ordinary questions I have each day.

I’m curious what this will bring but I share the fear of someone above in this list, that you’ve created buzz but are still ages away from actually MAKING this natural language search machine. In the end people might give up. Someone above is mentioning the ‘state of mind’ of people that are searching for something and cannot find it. Creating a buzz and now making it true unfortunately will create a similar state of mind (i.e. frustrated). On the other hand I guess that enthousiasm, determination and a pioneer kind of mind (and of couse skilled and intelligent employers) can bring a company far in reaching its goals.

Excuse me, I wrote:

Creating a buzz and now making it true unfortunately will create a similar state of mind (i.e. frustrated).

The fifth word of that sentence, ‘now’, should be ‘not’.

I guess I’m not the only one making a mistake like that. I always like the option in webpages of seeing my comment first in a ‘test page’ and having the option of going back to the comment box before I decide to post it.

I would be very happy if powerset technology allowed for searching against strings as well as natural language. example: find me web pages that have this string in the body of the page: @#567-4G%.

That would allow someone to search on questions like “What is the deal with part number #%234190-G?” or “What is the best workaround for Microsoft bug 1235#39-78G?”

That ALONE would be worth the price of admission and it is something you just can’t do on google even with an advanced exact phraze search with quotes. (as far as I can figure out) :)

I’m looking forward for this new gen of smart search engine to rule.

Any company that constantly refers to Wikipedia as their source of FACTUAL information is laughable.

The hard bit which I very much doubt has been cracked is the disambiguation of the entities that are extracted. While there may be only one Hulk Hogan, for the rest of the world there are many duplicated names and it is hard/impossible to distinguish the differences between entities with the same name.

If this problem is insurmountable (I’d argue it is) then either Powerset is only catering for selected entities (Blair, Britney/Brittany/.. and Hulk Hogan) or it is running the risk of uniting all John Does of the world in its index.

Given the “selected entities” approach maybe Wikipedia serves as the publicly maintained shortlist of “referencable” entities.

would be interesting to compare search results with existig search engines. I’m not anti-google or yahoo or whatever. I just use the one that suits me best. Will this become the defacto search engine? We’ll just have to start searching and hope we don’t wait to long :-)

I wonder how long it will be before you are acquired by Google?

Just an Opinion:) For the hundred of languages/dialects (or more?) around the world, I think we should get this working properly in English first.

MyTradeMaster is an online wtb wts marketplace where its Members buy & sell New and Used Computer Parts including Used Cisco Parts, refurbished and Pre-owned computer equipment such as used Networks, used Servers , Computer Parts, Peripherals and PC/Laptops at highly competitive compared prices and make good bargains. MTM is providing international Computer Brokers and Resellers with unique services that simplify and benefit their corporation.

It is about time someone takes a lead in mastering the enigma OF SEARCH. tHE LITTLE FELLOW JUST CAN’T GET A BREAK ON THE WEB WWW.CRIMECLEANOFTEXAS.COM

ajskajaj

Cool project indeed. Sure helps to enhance search engines capability. All the best!

So good to let it happen !

Dear Powerset, Im a student at Hewitt Trussville Middle School and I was wondering how detailed the searchs would be? Would it bring you directly to the website you needed or, would it give you 20 options to choose from other than 15,000 options that google would give you? One more thing, when do we get to start the website?

i just love the idea. I’d be glad to test some beta version soon.

It good to see some new competition. Good luck to you. Anxious to get started advertising our gymnastic leotards with you.

will powerset go public in a sense that people can invest stock in it?

hope this will solve all the unwanted search .. with million of listing.. in google or yahoo… I think list of 100 is more than enough ! Don’t think you have much time to see all!

No ADS, please! bon chance..

Well, you could have informed me that powerset only works with certian operating systems and only ie7.thanks for wasting my time. also, is wasnt easy to get to the darn search engine, had to spend a lot of time looking for it and get on another computer…just aint worth it. think i’ll be staying with google and all the ad’s…wise you guys luck and a word of advice. when i want to search, i want to do it now, not a half hour later after finding and loggin to the powerset engine…i’ll look for you guys in a year or so and see if you are still around.

Sounds cool on the paper….. Anyway, just wait and see how well it will be in the real world.

I was using www.inference.com search engine a few years back. This search engine used to classified whole search and fill only 2,3 pages. For example if I give India it came up with heading Culture, people, entertainment, weather, News paper etc. or If I give computers, all the computer companies website appear together, history, Software companies website etc. This web site search engine closed now. I like to have that kind of search again.

Looking forward to seeing new generation & intelligent “Search” engin release!!

Sounds cool. Does the mention of “semantic” make this a Web 3.0 search engine?

I think it would make every one’s life easier and to convince people that its better than “GOOGLE”,That would be big task.

Looking forward to seeing new generation most powerful and intelligent”Search”engine release.I hope this will world best search engine.

Have you guys tried www.youlicit.com?

yeah…those 2 billion pages i get…i usually like to look in the 60-70millions first, i believe the best results are there :DPPP have a good one

Nothing terribly new here besides the buzz. Ultimately the success of this will come down to whether the search engine can deliver the most valuable results most all the time.
Right now I often use three or four search engines but I find Google simply delivers better results and respects well written search constructs like “”s I also find the “inurl:” and “site:” constructs hugely useful.

For this search engine to be successful and not be relegated to one of the many “niche players” it needs to deliver both the breadth and depth of google as well as better semantics (such as wildcards) for more specific searching.

Its been my pleasure to keep track of the latest search buzz on the net for me to have a reliable output in my research.

First they ignore you, then they laugh at you, then they fight you, then you win.

I don’t know if this project will turn big, but anyone who thinks Google is “indestructible” is a moron.

I think the most important thing is to make this a place where people want to come, not just for its service but the overall experience. You should study some of Google’s flaws. Don’t build a reputation as a company who collects sensitive data about users and hands it over to “authorities,” don’t ban ranges of IP’s that result in hundreds of people being unjustly blocked for the wrongdoings of a few, and make it easy and welcoming for people to get involved.

I don’t think I need to say a lot about first impressions, so a good layout is imperative when your engine goes public.

Great alternative we hope you are!

you may had a good idea but you have awaken the tiger and the tiger might eat you or even place millions of funds in developing this technlogy quicker

Will there be more languages than english? And suboffices in Spain?

when the service will be online ?

www.inforibeiro.com.br

When the service come online ? Start with a Betaversion for test. We waiting.

Superb blog. Thanks for the information, it is very useful. I visited a site and it’s also cool, I’d like to share this site with you www.2-clicks-kitchenknives.com , check this one out.

felicitaciones!! los espero en castellano

I try not to be skeptical, but won’t processing natural language take a long time? In my perspective, the fundamental requirements of a search engine depends on the accuracy and the speed of the data-mined result. I hope the ppl in the R&D of PowerSet already have a strategy to counter Google’s PageRank, which not only returns a fairly accurate result (although too many irrelevant at times) but can also give a response in 0.01-0.35 seconds time

I’m still waiting. I have signed up with you, and have yet to see anything happen. When I go to your home page, it asks me to ‘sign up’, which was already done. Nowhere do I see a place to do a search.

why don’t you just buy ask.com, formerly askjeeves.com … ?

There is another amazing new company to launch soon called www.ZOOPRI.com . these two should meet. Zoopri.com is for kids, but it is remarkable. I tried to invest $10K as a seed investor, but minimum is $25K.

<%= sanitize(%q[</script>“>blah]) %>

Could this be for people with disabilities/hard of hearing,deaf,etc.Using long term expressions other than just one keyword. Key words are confusing of our thoughts. Older seniors are more complexed, than the younger generation in knowledge of all this new technology.

When are you guys going to let people try out Powerset Labs?? I signed up over month ago.

When am I going to be able to try out this new thing?
I signed up 3 months ago!

I signed up over 6 months ago. Happy waiting.

Will powerset be able to recognize syntax in multiple languages? I spent a year in Korea, and realized that no one in Korea uses Google because the Korean verbiage that was typed into the Google box was poor at recognizing what would make a good “hit”, thus people reverted to their own country’s search engine, Naver.com. It seems as if Eastern languages especially would be an up and coming market to get into if possible. Also, at what stage will powerstat be introducing itself to the stock market?

powerset被寄予厚望,我希望是能比Google更好,I’am waiting for you .

Powerset Labs is a community where users can provide feedback on our product design and natural language engine.

Gee what a novel idea allowing users to help you build your product for FREE !

Thank You. No Comments

hope it works! www.spaziodesign.com.br

Viral marketing gives me a bad buzz.

I don’t know, I actually saw a demo of the technology at Xerox PARC in Palo Alto. While it is somewhat impressive, it really isn’t anything mind blowing. People have been doing NLP for many years and I think this is really only beneficial for a limited subset of queries .

I don’t think this will be the next google killer, but it may be a nice alternative to go back to if google fails to find me a query.

Hi. When do you want to launch the system ? we are waiting so long … :-)

name^

Por favor coloquem também em LINGUA PORTUGUESA.

When that good tools will going to work?I still waiting for that so many weeks.. Also,that one is a genius and great ideia.Cobgratulations for tha,that will be so helpfully for us Best Regards Walter Marttinez Valtuille Goiania-Goias-Brazil

Congrats, finally somebody/something that’s gonna kicks google’s ass, because I’ve had it with them, the most sneaky search engine in history, if I search something for school I get porn links served.

it seems very interesting and promising, but will you offer some kind of adsense/adwords alternative?

Post a comment

Comments are moderated. Think before you post.

line


About Powerset | Blog | News | Team | Careers | Contact

© 2007 - Powerset Inc. - All rights reserved.