Session one as follows, including our live notes. Powerset Powerset is a natural language search engine that can use everyday phrases and grammer to conduct more accruate web searches by understanding the search query and the pages it indexes. Parsing phrases and grammer theoretically produces better results because the egine has a better understanding of the searches intended goal than with just keywords alone. For instance, a Powerset search for “politicians who died in office” returns information on the subset of politicians who died in office, rather than a group of pages that ranked highly with the phrase. Powerset presentation begins: talk about semantics and search, “we parse the web”. Natural language search. Announcement: Powerset labs, where users can explore tech demos, share ideas, feed the learning engine and “improve your search karma”. Demonstration of natural language queries with a social voting style feature. Touches of other sites Demonstration of Powermouse (see screen shot), information is pulled from Wikipedia into a semantic index. TC40 attendees will be amongst first in private beta. Overall: tough sell in the search vertical, but interesting take. Great start to TC40. Cognitive Code Cognitive Code makes artificially intelligent user interfaces. Their main product is the SILVIA (Symbolically Isolated, Linguistically Variable, Intelligence Algorithms) platform, which can add a human-like artificially intelligent interface to nearly any digital device. The SILVIA platform can learn and converse in natural language to carry out tasks for the user. Potential applications include children’s digital toys and personal assistants. Flagship product: “silvia platform” Symoblically isolated linguistically variable intelligence algorithm. Laymens terms: AI. Demonstration with AI on the screen, the AI system is having a conversation with one of the Cognitive Code. A couple of bugs in the live demo, but pretty cool. Uses include embedding in toys, phones, websites “unlimited uses.” First major target market is “smart toys.” Clever idea, if they can pull it off we’re seeing the future of toys. CastTV CastTV is trying to build one of the web’s best video search engines by creating a rich index of contextual data about videos and an easy to use interface for searching them. The engine pieces together context for a video based on it’s metadata, the content surrounding it, and the content of pages linking to the video. Notably, CastTV also searches paid video searches such as Apple iTunes. Their user interface allows users to sort results by shows (to → Read More
In a less than shining moment, Caitlin Upton, the 18 year old Miss South Carolina Teen, answered a fairly simple pagent question with a nonsensical answer: Q: Recent polls have shown a fifth of Americans can’t locate the United States on a world map. Who do you think this is? A: I personally believe that U.S. Americans are unable to do so because, uh, some people out there in our nation don’t have maps and, uh, I believe that our, uh, education like such as in, uh, South Africa and, uh, the Iraq and everywhere like such as, and I believe that they should, uh, our education over here in the U.S. should help the U.S., uh, should help South Africa and should help Iraq and the Asian countries, so we will be able to build up our future for our children. Not one to miss a PR opportunity, yet-to-launch natural language search engine Powerset took a shot at parsing her answer so that queries could be run against it. Based on the query “Who does education help?” the index returned the result “Americans.” That’s an impressive result, given the nature of the data being queried. The test shows the potential usefulness of Powerset as a search engine. The query does not match the content based on a keyword match, and the answer can only be determined via a contextual analysis of the data. Powerset tends to look very good in demos against a limited index, as the above example shows. but it still has to prove that it can index and analyze large chunks of the web to become a viable competitor to Google and other search engines. That’s going to be their biggest challenge (and cost). Powerset still has much to prove as they prepare to launch. → Read More
New natural language search engine Powerset, still in pre-launch stealth mode, has had a ridiculous amount of press this year. And while some have said there is too much hype around this company (even me), you have to give them some credit. They are certainly open with their plans, and willing to experiment with new ideas. An example: they announced Powerlabs, a sandbox for users to suggest and give feedback on future Powerset features. People who sign up for Powerlabs are also promised early news, at least an hour before it is posted on the Powerset blog. Another example: In May Powerset COO Steve Newcomb talked about how the company was predicting future growth, and posted data on their model on the company blog. When readers bravely requested that Powerset release the model itself, Newcomb complied, saying it would be made available this summer. In a post on his personal blog he said the reason for sharing the models was to show that the company intends to be open and give users unfettered access to information: As I mentioned before, opening up our modeling techniques is part of a larger goal to begin the process of changing our image of a secretive stealth startup to a completely open company that gives you unfettered access to our product(s), the ability to help us design them and to provide insight into the way we think inside of Powerset. Today, Powerset published the first in a series of models, with a Flash interface. Company-specific baseline assumptions have been removed or altered, but most of the industry assumptions remain intact. Neal Mueller (Powerset Product Manager) walked me through the models and how they work. This first set helps a company that intends to index the web whether it is better to purchase, lease or create virtual servers on Amazon EC2. Assumptions about the size and refresh frequency of the index can be changed. Since the model is forward looking, it also makes assumptions about future server power and cost reductions from Moore’s Law. All of the assumptions can be altered in the Flash interface, and the models can be embedded into other websites (although I could not get it to properly embed here). Mueller says that at least two more dashboard models are coming – one for unique user forecasting and another one that they are not yet disclosing. The company is asking for → Read More
Xerox announced its entry into the search market this week with FactSpotter, document search software that is claimed to go beyond conventional keyword search. FactSpotter is text mining software that combines a linguistic engine that allows users to make queries in everyday language. FactSpotter looks for the keywords contained in a query along with the context those words have. According to Xerox, FactSpotter is capable of combing through almost any document regardless of the language, location, format or type; take advantage of the way humans think, speak and ask questions; and discriminate the results highlighting just a handful of relevant answers instead of returning thousands of unrelated responses. Frédérique Segond, manager of parsing and semantics research at XRCE said that the tool is more accurate because it delves into documents, extracting the concepts and the relationships among them. “By understanding the context, it returns the right information to the searcher, and it even highlights the exact location of the answer within the document”. Whilst it sounds appealing, FactSpotter will not be coming to a browser near anyone, anytime shortly. Xerox plans to launch FactSpotter next year as part of the paid Xerox Litigation Service platform and has no plans for a wider or public release. Here’s betting that a Steve Jobs character comes along and steals the concept and turns into the next Google; history often does repeat itself. → Read More
When Powerset announced plans to launch a new service called Powerlabs a week ago, it looked to be another sand box area for product previews, like those created by Microsoft, Google and others.But today Steve Newcomb, COO of Powerset, revealed a lot more about the project. I also saw an in person demo of Powerlabs today and some of the ideas behind it. Powerlabs is more than a sandbox to show off new product ideas that aren’t ready for prime time. They are encouraging people interested in Powerset to sign up for Powerlabs and create what is effectively a profile. Once registered, users will be able to see new product ideas and vote on them, as well as submit their own ideas to the community. Later on, users will gain points and influence within the community. In the demo today the company showed me one idea that they will be putting into Powerlabs once it launches – a mashup of Powerset natural language results along with video results, keyword results and, interestingly, Freeweb results (the two companies are working together). See the screen shot to the right (click for larger view). Powerlabs hasn’t launched yet, but you can register for it via email on the Powerset home page. They are sending out regular communications to those users, including news at least an hour before its posted on the Powerset blog or given to press. In an email today, Powerset revealed that they have acquired another company, although they did not give any further details. → Read More
In a post last month I wrote about Powerset COO Steve Newcomb’s use of predictive modeling to guess early growth rates so that they have enough hardware to scale. Good for them, I said, for releasing some of the data publicly. And I recommended they go one step further and release the predictive models themselves: Powerset should publish the model itself (without the specific Powerset assumptions of course) and let other startups tweak it for their own use. Most new companies don’t have the excel jockeys or the time to do this kind of work. Any competitive issues would be overshadowed by the considerable goodwill (and link juice) they’d get from doing this. And that’s exactly what they are going to do. Newcomb says they’ll release the models in the next few weeks, minus Powerset-specific data. Don’t expect to see these on Google Docs, though (a perfect place to dump them, except that Google is who Powerset is gunning for). They’ll make the spreadsheets available for download or let people view them on the web via a Flash viewer. → Read More
Powerset is being extremely careful about showing the public how their search engine works until they are ready. After some initial hype (see our posts here, here and here), the company pretty much shut its doors to the press. I did finally get in to see a demo, and was impressed. But the meeting was off-record and we are waiting for a green light to start writing more about the demo and other background information. For those of you unfamiliar with Powerset, it is a new, well funded search engine that aims to allow users to write their queries in natural language. In a blog post in October 2006, CEO Barney Pell wrote out some of the ideas driving the company. A couple of days ago the company showed a sample query and result on their blog. Normally boring stuff, but a lot of people are dying to get more information on the product. We’ve copied the screen shot above. The query is “politicians who died in office.” This is a much better query that previous examples like “books by children” v. “books for children.” Those queries can be handled fairly well by Google by simply putting quotes around the query. But for “politicians who died in office” the results on Google won’t be as good. Context is required: Google has only six results for the query in quotes, and without quotes it loses its meaning and the results aren’t useful (notice the Powerset blog is the fourth result). The Powerset results are relevant and useful. Hand picking a query here and there and showing a screen shot of results isn’t the same as killing Google. But it does show that Powerset has the potential of being extremely useful by attacking search from a different angle. I look forward to their launch. → Read More
Any serious Internet startup tries to guess early growth rates so that they have enough hardware to scale (hopefully the software hangs in there, too). Stealth search engine Powerset (see our earlier coverage) is being very transparent about their modelling, and have posted some of their thinking on their company blog. Steve Newcomb, co-founder and COO, wrote the post and goes into considerable detail as to how they built the predictive model. The graph above shows Newcomb’s model as applied to a competitor – Snap.com, after the new Snap Preview Anywhere release and its re-launch of its product. The shaded area shows predicted growth and the actual results are shown by the blue line. Not a bad job, although I assume they tweaked the model quite a bit to make it fit. Powerset should publish the model itself (without the specific Powerset assumptions of course) and let other startups tweak it for their own use. Most new companies don’t have the excel jockeys or the time to do this kind of work. Any competitive issues would be overshadowed by the considerable goodwill (and link juice) they’d get from doing this. → Read More
The more I dig into the new search engine startup Powerset the more I am wondering if it is nothing more than a house of cards. They’re an odd company with a bit of a split personality. For example, in some ways they are very secretive – everyone who gets to see “The Demo” (as it’s now being called) have to sign a nondisclosure agreement. It’s rare for startups to request this. On the other hand, though, CEO Barney Pell is hyping the hell out of the product (which is still just The Demo – Powerset may or may not ship in 2007 from what we’re hearing), the fact that they’ve raised $12.5 million in venture capital and that they’ve hired a dozen or so search experts out of Yahoo (where options are vesting and the search experts are itching for the next big thing). All this hype is generating a lot of press. Normally even handed Matt Marshall continues to gush about them every time he writes (he got the chance to see The Demo). And the New York Times has written about them twice already this year (here and here). This does make a good story. People are frustrated with the current state of search and want something better. Google was also criticized when it first launched as trying to tackle a problem too difficult to solve. People draw an analogy to Powerset, perhaps thinking that once a decade a new company will launch and kill off the old, entrenched competition. Still, Powerset is given little chance by the search engine elite who judge such things. People who’ve seen The Demo say it is a completely controlled environment. The index is limited to just a very small sample of high quality sites like the New York Times website, and the search queries are driven by Powerset employees as well. With that kind of setup, almost anyone could show a stunning demo. The real experts I’ve talked to who’ve seen the demo said it shows little or nothing real. Others leave impressed. The company had a party in San Francisco this last weekend to celebrate their venture funding. Sarah Meyers, who regularly crashes parties with a cameraman to videotape what’s going on (including ours), was there to talk to Pell and other employees. The video is embedded below. Pell talks at a high level about the product, not really → Read More
Silicon Valley based search engine startup Powerset has mostly been closed lipped about their product. This makes sense given that they are gunning for the fastest growing Internet company in history, Google. But their excitement over an exclusive deal to license PARC search technology was too much to hold in – the company made what looks to be an exclusive announcement through VentureBeat today about the arrangement. Powerset argues that the key to killing Google is in natural language search. We discussed this in detail in an earlier post about the company. While other search engines tend to ignore common words in search queries (words like “by”, “for”, “about”, “of”, and “in”), Powerset focuses on those words to try to determine meaning and context. Read CEO Barney Pell’s post about the natural search problem to get an idea of what they’re talking about. To get right to the point: Google treats the queries “book for children,” “book by children,” and “book about children” as equivalent to “book children.” Powerset, however, promises to serve results that are relevant to each query. There are a lot of critics of Powerset, who note for starters that Google will treat each of the above queries differently if user simply put them in quotations. Search engine expert Danny Sullivan took the Powerset idea apart piece by piece in a long article last year, saying “natural language search makes a compelling pitch for those who really don’t know search or haven’t heard the natural language mantra before.” Search experts we’ve talked with about Powerset agree to varying degrees – the problem is very, very hard to solve. And users have largely learned to simply change the way they search to get what they are looking for. It’s impossible to judge Powerset before we actually see the product. They are, however, trying to solve a very difficult problem for which there may or may not be not much of a market (at least compared to simple, 2-3 word searches). And Google has “several teams focused on natural language and dozens of Googlers with a PhD in the field, including myself,” says Peter Norvig, Director of Research, Google. If we start to see some of those PhDs leave Google and join Powerset, that will be the first sign that the company is really on to something big. Powerset is well funded, having raised $12.5 million from Foundation Capital, → Read More
Update: I received an email from a trusted source this morning that reads “…I’ve seen the demo (and spent a fair amount of time with those guys)…The technology is quite good and will certainly help on a non-trivial subset of web search queries. But they definitely have their work cut out for them. The indexing is far more expensive than traditional web search, and that means some real scaling and performance work to do…Can’t say much more without really pissing off the NDA gods, of course…(Oh, and those VentureBeat numbers are wrong about their funding.)” I haven’t seen a demo of upcoming search engine Powerset yet, but reportedly many people who have are impressed, saying they’d never use Google again. Powerset wants to let people use natural language when searching, including some words that search engines ignore today (what founder Barney Pell calls “stopwords“). Like most people, I’ve learned to change my language when addressing a search engine. I use important keywords and leave everything else out. Some of these search habits have even overflowed into my personal communications, and I find quick IM and email discussions often look like crazy, meaningless words thrown back and forth. Ten years ago I’d have no idea what those discussions meant. Today, they are effective in getting through fast paced business conversations with people I’m already familiar with. So while I believe that search is far from perfect today I didn’t necessarily think that the solution was to create a better way for computers to understand what we meant. Frankly, I assumed that we’d continue to do a better job of talking in a way that computers understood, and advances would come in other areas (deep web, better algorithms, rich media search, etc.). But Pell lays out a convincing argument that natural language search is important in order to communicate meaning and intent. He uses example searches to make his point – “book for children”, “book by children”, and “book about children” are all equivalent to “book children” to search engines today. His core argument is that there may be no way for us to properly express the query “books by children” without using natural language. Powerset is looking for big money to launch their new engine. Venture capitalists are always the best source of rumors (the best time to hit them up for information is right after they’ve passed on a deal, → Read More
San Francisco, CA
