Microsoft will announce today that they have acquired San Francisco based semantic search engine Powerset. The acquisition price is not being disclosed, but our understanding from sources close to the deal is that the previously rumored $100 million is “roughly accurate.”
In May we reported that Powerset was in acquisition discussions with Microsoft and was hoping to bring another bidder to the table. Google was the likely candidate, but they have publicly dismissed the notion of contextual search as a revolutionary step forward. Microsoft, which is clearly interested in improving its search market share, turned out to be the best fit.
Rumors resurfaced last week about the imminent deal.
Powerset recently launched a showcase for its semantic search product, although they lacked the funds to do a full web index to prove out the product. As part of Microsoft, they won’t have that problem any longer. Now they just have to fight the bureaucracy to make sure the project continues to move forward.
The company had raised $12.5 million in venture financing, plus another $8 million or so in convertible debt as bridge financing. That means investors will get a decent return (but not a home run), and the founders and employees will also take some real money off the table.
We first covered Powerset in October 2006, and they were a TechCrunch40 company.
Great! One less company for Google to buy up!
Google’s search seems to be working OK without semantic search. Semantic search might be an interesting science fair project but I’m yet to be convinced that it will make the world a better place.
Will be interested to see if they can make an order of magnitude increase in relevancy in results with Microsoft’s support. Google may have this locked up until the next real paradigm shift unfortunately…
Ha, you didn’t get the story first. I can tell you’re still butthurt over VentureBeat breaking it.
anybody else realise that Powerset alongside FAST is just another move by Microsoft to steal Google’s search share? Either that or they’re looking at enterprise search - possibly an area that they’d have better reception in since they have so much enterprise software already and their web search isn’t anywhere near used as much as Google.
Either that or it’s a plugin for Live Messenger… “Find me a restaurant near New York” etc
MSFT is wise IMO to spend a fraction of the Yahoo bucks on a set of acquisitions in the structured search, semantic search, and Internet-as-an-operating-system areas. They also need to clean up the brand confusion around “Live” and what it means.
Google has done many things right, but is vulnerable with the advent of the structured web and potential leaps forward in semantic search.
I say potential, because right now, I’m not seeing Powerset’s results as revolutionary, but I think they’ve done deeper thinking on this than most other semantic search companies and the addition of the team could really help push MSFT further.
(am a former MSFT employee)
awesome. live search should progress with this and powerset now has the resources to move ahead and do greater things.
With Microsoft on board, Powerset will definitely have the financial capacity and resources to make it a full blown web search.
Let’s hope they will do it right this time.
Looking forward for great results!
Best regards,
Darren Lee
http://www.adexcel.com
Powerset is a big ruby and rails house. They also have made extensive use of cloud computing and ec2. I wonder how the engineering culture is going to fit in to MSFT.
Venturebeat broke this. I am switching blogs. Too much ‘I did this first’.
It’ll be interesting to see if Powerset will still continue contributing to open source projects that it heavily relies on, especially Hadoop, which Yahoo! maintains and uses.
Powerset’s blog doesn’t mention anything about this.
Congrats to the whole Powerset team - Microsoft will be a great platform to leverage, and the fact that they are much more committed to a semantic path than GOOG makes them a much better home for the team.
On the open source, et al question, MSFT is past that being a religious battle. Look at their $6B aQuantive acquisition as an example of another big deal for a business heavily employing open source.
It will be interesting to see if Powerset will remain committed to Hadoop and/or Hbase after the deal closes.
Michael, your terminology is a little off in the article.
Powerset isn’t Contextual Search.
Contextual Search is when you find information related to the users context. Powerset is Natural Language Search.
E.g: “ThoughtTrail has Contextual Search”, or “Google Ads use Contextual Search”.
There terms are not mutually exclusive though. E.g: ThoughtTrail has Contextual Search as well as some Natural Language Search that does deeper understanding.
stupid microsoft buying unproven technology. Now they want people to write lengthy queries for search. Sorry Google keyword search rocks for me and save time (even if it is more than one search attemts). Even the long sentence queries work great on Google. But then Microsoft is “Microsoft”.
Few companies if any are making any money in the semantic space. Powerset was an acquisition of a team, but not a real business. The only company I know of that’s successful (i.e. growing rapidly and very profitable) in the semantic space is Relevad (www.relevad.com). Relevad is a semantic platform with multiple services for advertisers and publishers.
Powerset is the best looking search engine out there - IMHO. Also they have nice UX features implemented and they cover a niche section - Wikipedia and Freebase.
Microsoft saw the potential in Powerset’s approach and I think that it is good news for investors and founders. Also I think that Microsoft should continue buy Xonbi and the other mail collaboration service Orgoo in order to make a stand in the communication area.
Anyway - KUDOS for Powerset
Can anyone explain what Microsoft actually gets for their $100m ? From using their site, I can’t see any indication that what they’ve done is better than something two people could build on top of solr in a weekend.
It looks like they’ve hired some great people; is this simply a way for Microsoft to buy a team ?
Microsoft are buying this so they can integrate it with thier voice recognition technology. Don’t forget Gates is still there one day per week, he’s really into voice and MS have spent a load of cash developing it.
Natural language search is a much better fit for voice recognition than text search.
Don’t you think Google figured this out them selfs already but keeping it under the hood until it’s really needed?
So now all the folks at Redmond need to do is figure out how to snatch up the 30 odd % of YHOO to give Powerset a decent sized playground.
FAST is a great search engine, just as good as Google and Yahoo’s Inktomi tech, tech-wise, but the user isn’t going to see a big enough difference the switch. PowerSet adds that nice human “aha” moment that will get the user to get them to try it. It’s mostly a sugar coating, but that’s a big part of what MS needs if they’re going to run a Coke vs. Pepsi war with Google.
Nobody seems to be aware of the fact that all of Powerset’s semantic technology was licensed from Xerox.
Oops.
This is excellent news; semantic search is a great, compelling technology– and Microsoft has the funds and brainpower to help Powerset realize its full potential.
An earlier commentator, asserting that Google will continue to reign supreme until the next paradigm shift in search, appears to be missing the point. Semantic searching IS the big paradigm shift.
Sure, keyword search is functional– but it’s strictly a heuristic. When you ask a computer to find a piece of information for you, that computer needs to either (1) have sufficient predefined metadata with respect to both its information index and the user’s query so as to derive a quantitatively semioptimal result set, (2) attempt to generate that meta data on the fly using yet another heuristic such as PageRank [a la Google], or (3) go the distance and actually build software that doesn’t need no stinkin’ metadata and can fully understand the information it catalogues and the queries it processes so as to actually return results that don’t just seem relevant according to some arbitrary metric but that are truly and significantly meaningful.
Now, I’m not saying that Powerset is at (3) just yet; I’m just happy to see that semantic search is finally getting some serious backing.
#18 (Robleh) has it right. Powerset’s natural language technology is most useful when integrated with the voice recognition capability with Microsoft’s TellMe acquisition. I would expect MS to weave voice recognition/search capability throughout its product line.
The company hasn’t built any business model nor demonstrate anything that’s usable to consumers. Semantic that requires more information from me. Not information from a word or a phrase. Semantic means, just that, meaning that is meaningful to me, not to anyone else. To determine that, you have to collect as much information about me as possible. That’s how you can create meaningful results. That is semantic.
Semantic based on collective knowledge is not. Collective knowledge and making want is relevant to MOST people is Google. But what is meaningful is what I know and do.
I don’t think this acquisition will go anywhere.
Well done to the Powerset team.
Harold Carr said…
Powerset’s natural language technology is most useful when integrated with the voice recognition capability with Microsoft’s TellMe acquisition.
I agree, the downside to NLP is the longer sentences of a query which most people have to type and this is where numeric-based algorithm excels. NLP search is suited to voice recognition , but that is the future. Numeric-based algorithms are still superior than NLP, however we will see over time that the two will merge and that will be the killing search app of the future.
Cool, why don’t they buy the internet while they’re at it!
So Powerset is for people who have never used the internet before?
Daren Lee said…
Powerset will definitely have the financial capacity and resources to make it a full blown web search.
It would be slow and the relevancy drops. Web search works on links and NLP works on words, it is still better to use the link-search (numeric) such as PageRank as its primary engine, then use NLP as a secondary engine just to refine it further, ie, link-search retrieval result is feed-in to an NLP engine to refine it further & increase the relevancy.
If the $100 million purchase price is true, big congratulations to the founders. In real life that translates to a $100 m payday for simply securing exclusive license of the Xerox technology.
So here’s an idea, instead of building a search engine out of PowerSet, use PowerSet to search inside results already delivered by a search engine (but do this on the server side). So basically if I’m searching for “What things did Thomas Edison Invent”, I would enter that into the search engine (live.com?) and what live would do is pull up every page that matches those main keywords, then, before giving me the results, runs a powerset search on THOSE results to provide a semantic answer. Would not have to scale up Powerset beyond is current capabilities much, and would be a consumer side product within a month. Thoughts?
Powerset’s Natural Language search is not about applying NLP to the search terms. Rather, it is about applying that technology to all the web pages in advance, and adding metadata to each page. It takes a tremendous amount of computes and storage to do this for the whole web. But, once it is done the results are truly amazing.
Powerset did license some technology from Xerox PARC, but Powerset has built most of the technology themselves.
Lastly, Powerset’s technology can be used in web search, enterprise search, voice search, and many other applications. Powerset is not just about competing with Google. There is a lot of money to be made in other markets.
Common Sense Location will beat a natural language search everytime. Dearch is blind and kephrase vertical domain location is pinpoint and common sense.
When you have common sense keyphrase vertical location on a 1200+ multi-channel natural language locator network you dont need anything else.
Natural lanugage keyphrase location is king and who ever has the greatest multi-channel natural language location network wins.
what could possibly take all these self professed search geniuses so long to figure all this out. living in a bubble? pure posturing by mircro, have to look hungry and not broken hearted and confused. they have a clue as to the future of search problem is, its just that they just dont have the right answer.
AdviceLocator.com says get out the bubble and step into the real world. In order to be a real player you must possess or attain some or all of the following attributes:
Create a multichannel strategic niche market location engine network for every subject . Offer businesses and consumers common sense, strategic, keyphrase, location based, top level, vertical, multichannel, social, personal, custom niche offerings.
Learn about “Natural Language Location” @ LanguageLocator.com
I hate to say it, but even if Microsoft uses the Powerset acquisition to create a better search engine than Google, will it matter? I still don’t think people will switch. Unfortunately, Microsoft is just too despised a brand. Why do you think that Linux and Firefox have gotten so much traction in the open source community (hint: it’s all MSFT hatred. okay that wasn’t a hint, that was just giving the answer).
MSFT would be much better off just trying to invest in creating greater earnings per click. That means continuing to buy up inventory and technologies that can help improve ad targeting/relevance. After all, the future of software isn’t about *search*, it’s about ad-based business models. MSFT doesn’t need to have the dominant search engine. All they need is a healthy ad network.
My be, MS will not need the Yahoo after all…
@18 Robleh - made a good point. NLP is more suited for voice search. Yes, makes sense. One wise comment. Thanks.
This is FANTASTIC NEWS.
Powerset is a waste of electricity. Hypeware. Junk. I hope Microsoft keeps on buying this type of crap for huge sums for a long time.
I predicted that powerset was going to the deadpool, and they went to microsoft. Money issues aside, I wonder whether there is any difference. This type of tech is still at least 10 years away, and they’re not going to develop it with the wrong philosophical foundations.
I think this is a good acquisition for Microsoft; however the issues are larger than the under engine itself. Usability and trust are becoming a big issue for the software giant. The variations in search results between Live.com and MSN Search, plus the constant redirects to each, get very cumbersome for users. I won’t even get started on the AdCenter PPC platform, but I have made my thoughts know to their feedback forms. I think the best thing Microsoft could do with its search engine is to get the community involved. What are users missing and what are professional search marketers looking for.
This aquisition could be interpreted as more evidence that the semantic web is gaining favour amongst the big guns (Yahoo are apparently embracing it too).
There’s so much information on the web now, that we need to start giving it context in order to make sense of it.
http://blog.swirrl.com/article.....-important