No venture capitalist’s week is complete without seeing a company in the semantic space holding steadfast to the assertion that its pre-money Round A valuation is a steal at $50 million. The situation wouldn’t be totally absurd if it were not for the facts that these technologies are mostly half-baked, or far from becoming a deployable product. And they are most definitely leaps and bounds from being able to generate revenue—let alone actually generating revenue at the time of their pre-money $50 million valuation pitch.

But in the middle of this circus is a modest and press-shy company called Peer39. The company, which has purposefully been under the publicity radar since its inception in 2006, is a promising example that semantic technologies have true monetization potential. Peer39 is demonstrating this thoroughly through semantic advertising, the next evolutionary step beyond behavioral advertising, which was preceded of course by contextual advertising.

Peer39 is spearheading this space with its semantic advertising network, which is on track to hit one billion impressions by the end of the year. Their network only handles display ads at the moment, charging advertisers based on a CPM (cost per thousand) or CPC (cost per click) basis.

In an interview with TechCrunch, Amiad Solomon, the company’s CEO, claims that through the networks’ 30 top-level categories and thousands of sub-categories, Peer39 is providing advertisers, on average, clickthrough-rate improvements of four times what they were getting before. Publishers hail from all verticals, including technology, automotive, finance, and health.

Powering Peer39’s ad network is SemanticMatch, an algorithmic engine able to perform precise matching of commercial offers with user-generated content content (blogs, forums, etc.). SemanticMatch employs natural language processing and machine learning to determine meanings, topics, categories and even sentiment. All this, mind you, is done in real-time and Peer39 does not employ cookies in any way, nipping many privacy-related concerns in the bud.

Peer39’s technology has been two and a half years in the making. The 44-strong Israeli R&D team includes the former VP of R&D at Nielsen Online, as well as scientists from the Technion Institute (Israel’s MIT) and the Israeli army’s prestigious intelligence units. Together they were able to tackle the challenges of breaking down text into small enough pieces to categorize text with algorithms that are 1) generic enough to be applicable across a wide variety of content categories, and 2) able to adapt themselves to new categories on-the-fly.

Unlike contextual analysis that seeks out key words, Peer39’s semantic analysis engine—SemanticMatch—looks at the entire page to derive its meaning and its relevant categories. Below are a couple of examples provided by the company (I’ve added the italics for emphasis):

Example 1: “Amazon is a great place to visit.”

Peer39’s Category Analysis: Travel, Brazil, South America.

Example 2: “Amazon is a great site to visit.”

Peer39’s Category Analysis: Ecommerce, Internet, Business and Finance.

Category granularity can go down to seven levels. For example:

Technology > Consumer Electronics > Digital Camera > Canon > Powershot > SD870 IS > SD870 IS Silver

Auto > Compact > GM > Chevrolet > Chevrolet Cobalt > Chevrolet Cobalt LS Sedan

Peer39 is also able to determine positive/negative sentiment. This is important because it allows Peer39 to offer its advertisers various brand protection thresholds, such as offensive content, negative content, mention of competitors, etc. Canon, for example, might not want to advertise on a page with a negative review of one of its cameras.

On the business execution level Peer39 is attempting to pull off a Tacoda or a Quigo exit. Both companies were able to develop sizeable ad networks by delivering quality ROI results and working closely with agencies.

The company has raised $11.7 million from Canaan Venture Partners, Dawntreader Ventures, and JP Morgan. (The company won’t disclose its pre-money valuation, but given how much it has raised, a valuation north of $50 million is a safe bet). Its board of advisers includes Eytan Elbaz, a co-founder of Applied Semantics (which was bought by Google and became the basis for AdSense) and the former president of Tacoda Daniel Jaye. Also, the company just announced that Matthew Goldestein, former SVP of Revenue Operations at Tacoda joined as COO.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

Meet Jogli, a music search engine that claims to offer immediate listening access to 500M songs and 12M albums.

It would be valid to argue that there can’t possibly be more room for yet another music destination. However, having spent the past few days using Jogli, I have to conclude that it strikes me as having the potential to shake-up competitors including Songza, MeeMix, Deezer, SeeqPod, and a number of others.

Jogli crawls the web for music and music clips and then indexes them for search (the majority of songs come from YouTube, but the site will crawl other services). Jogli then lets users listen to the music it has discovered through a player integrated directly within the Web interface. Sure, “search, click & play” is nothing new, but applying it on 500M songs is a significant feat.

From my personal testing Jogli provides effective music search for artists, songs, whole albums, and clips. I easily found every non-mainstream artist or song I searched for—examples: The song ‘Thin Line’ by Jurassic 5 featuring Nelly Furtado, and Nick Cave’s album ‘Let Love In’.

The trend of piggy-backing YouTube via its API is changing the rules of the game for music services. Remember, Internet radio services like MeeMix and Last.fm have to pay royalties, while Jogli pays none. They don’t even pay streaming or storage costs—YouTube (meaning Google) foots the bills. Of course, this trend may not last long if YouTube starts cracking down on these music videos.

Jogli is in Beta so it’s far from perfect. Community and discovery features, for example, are practically non-existent. However, an inventory of 500M songs is surely enough to get Jogli foot in the door.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

For the benefit of TechCrunch’s American readers, let’s first set a basic ground rule for this post: Football = Soccer. You know, the game played 90 minutes that can end in a 0-0 draw but can still be regarded as an excellent match. With that out of the way…

Second to the iPhone 2.0 announcement (and maybe the Yahoo-Google deal), what was the most important event of last week? The start of Euro 2008 tournament of course! So if you’re going to soft-launch a football oriented social network like Footbo.com, this is the month to do it (the tournament runs throughout June).

Footbo is the type of niche-oriented social network we have grown accustomed to. From a feature perspective there’s no attempt to reinvent the social network experience—in fact, the current interface doesn’t offer any Ajax/DHTML/CSS pizzazz. The idea of “less is more” rings true in Footbo’s case, which focuses on offering an information and interaction hub for everyday football fans, as opposed to cutting-edge functionality and interface design. Even so, the design is at times rough around the edges with tables running amuck and such.

Footbo provides football-related content in the form of live feeds, line-ups, game schedules and tables from about 70 leagues around the world. A Predictions section allows users to predict the final scores of games. These can be submitted/edited until the kick-off of each match. Users can also grade player performance, thereby electing the “Top 11? of each league or round. Integrated wiki functionality lets users add and edit content about teams and players. There are already tens of thousands of existing profiles with basic info on teams and players. Game pages feature fan boards and live discussions. The Euro 2008 section leverages these features on a single page for a “one-stop” experience.

Footbo is predominantly a European play targeting individual fans, fan clubs, and amateur teams. For the sake of context, Footbo estimates there are 170,000 amateur football teams in Germany alone. Properly executed, there could be plenty of business servicing this niche alone. However, Footbo’s scope is wider as is evident by the fact that the site is available in English, German and Spanish.

Footbo will launch its official Beta in August with a host of new features. It’s up against some serious competition from ESPN’s Soccernet and less serious competition from Joga.com (a soccer social network that was created as a joint venture between Nike and Google’s Orkut which is now seemingly defunct—all that is on the site is a splash page promising that something called JogaTV will be “coming soon”).

North American readers: You can now revert back to the term “Football” referring to a game with far too many rules played with an egg-shaped ball.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

It’s a Wii without the $250 console. It’s virtual Pong and so much more. Any object is now an input device, even your fingers. Bang, bang! But there’s no better way to introduce you to CamSpace than by letting you watch the demo video below:

CamTrax’s core technology is a pure software solution that allows nearly any ordinary PC webcam (95% are supported) to track up to four objects—even as small as 5mm—in real-time and with very high accuracy and reliability. (It works only on Windows). Locking and tracking (X, Y, and Z axes and angle) are all automatic. Yaron Tanne, founder & CEO of CamTrax Technologies, the company behind CamSpace, has been developing the technology practically single-handedly for three years in his apartment in Tel-Aviv.Tanne claims that most of the algorithms used are in the public domain but have been enhanced. There are also completely new algorithms developed from scratch.

CamSpace requires an agent application to run locally in order to emulate a mouse, a keyboard, joystick, or other input device. Users can then program the emulation based on the game they want to control and the object(s) they want to control the game with. For example, one user could program a steering wheel for a racing game, where moving the wheel on the Z axis shifts the gears up and down. A different user can use two objects for the same game, programming the second object, say a coke bottle, to shift the gears.

Once a good portion of most popular games are emulated, the company will provide a portal where these emulations will be rated based on popularity and then offered for download.

Assuming there are no patent infringement issues, CamTrax could be a hit in several sectors, the most obvious one being gaming. While certainly the big game studios could take advantage of the technology, I can see a wider and quicker adoption among casual gaming entities such as Zynga and SGN. Cam-Trax could also find success in providing solutions for handicap individuals that cannot use standard input devices. Another application would be to emulate multi-touch control over media-centers. Some more ideas:

• Fitness programs using body movements
• Virtual instruments (air drums, xylophone, etc.)
• Drawing “in the air” applications for kid

These are just the tip of the iceberg… Remember, all you need is a standard webcam—that’s a VERY low barrier to entry these days.I’ve had the chance to play with CamTrax’s technology on several occasions and it works like a charm. This is true even in low and changing light conditions—based on first-hand experience. It really is hard not to be impressed with the technology, especially seeing as it still has Alpha status.

The four-man team recently raised $200,000 in seed financing from angel investors and plans on raising a Series A round in the coming months.

In the meantime, the company is working on a developer platform which will allow the integration of CamTrax technology into casual games and mini-applications. Expect a follow-up post when this happens in the coming weeks.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

Last week The Globes’ Noa Parag interviewed ex-Yahoo CEO Terry Semel (he resigned in June 2007) Shimon Peres’ “President’s Conference” in Israel. Also on the panel was Sergey Brin, Susan Decker, Rupert Murdoch and Yossi Vardi.

Parag dug to try to get something juicy out of Semel on the Microsoft/Yahoo battle (this was right as Carl Icahn was making his effort to replace the Yahoo board public). The original interview is in Hebrew, but here are some of the translated highlights:

Parag: But you personally believe that Yahoo is worth more than $44B?

Semel: That question should be directed at Yahoo’s new managers. To Jerry Yang. They seem to think it’s worth more.

Parag: What’s your opinion about Icahn? It is an attempt to coup the Board of Directors?

Semel: I read about that as well. Icahn is very talented and intelligent, but I don’t know what intentions stand behind the stock buying.

At this point Semel requests to stop the interview for a moment. All these questions about Microsoft are making him feel uncomfortable, and he announces: “I’m not going to talk about the deal. Who should be talking about it is Yang or Susan Decker, the president. Certainly not me.”

Semel also says his departure was based on personal reasons:

Semel: When I began working in Yahoo, my family moved with me. Despite our efforts, our kids wanted to study in Los Angele, and I was forced to see my family and friends only on weekends. In the beginning I even enjoyed it, but knew that at some stage I’d want to go back home. At the time, the people that founded it wanted to get back to it, and that was the right opportunity to go back home and look for new opportunities.

On speculation that his exit was forced:

Semel: These rumors are not true. I told the Board several months in advance that I was going to announce my departure soon, and that I want to set out in a new direction. They offered that I stay as Chairman, and I did agree for a certain time, but at the end felt that I prefer to be an active player than a mentor. It was the right time for me to leave.

On Facebook’s valuation:

Parag: Is Facebook in your opinion worth $15 billion? Are these prices even logical?

Semel: I have no opinion about it. This is something Microsoft needs to decide on.

On future plans:

Parag: So what is Semel doing these days?

Semel: I’m busy mostly with in searching for interesting companies in order to purchase and invest in, and I’m certain that they’ll be heard of in the coming years. I also have time to play golf and do some sports.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

MeeMix keeps plugging along on its journey to attract more users to its personalized Internet radio service. Today the company is adding a new feature which complements users’ radio stations with “twin” music video channels.

While Last.fm introduced video recommendations exactly a year ago, MeeMix thinks this will help make its service more attractive. The company has always felt its taste prediction technology is not strictly limited to audio, so from its perspective offering taste predictions for video is somewhat of a logical next step.

Enabling MeeMix’s video mode requires no setup at all. The feature leans on the taste selections existing users have already made when setting up their existing radio stations (i.e., music genres, artists and songs). Relevant music videos are then fetched from YouTube. Editing radio stations, for example adding a new artist, impacts the video selection immediately.

MeeMix fetches a wide range of videos—from official music videos, to very amateur material. Some users will find this quite entertaining. I for example had to stop writing this post mid-way to enjoy a vintage video of John Coltrane playing Naima.

The video sound quality falls below the audio stations’ 128 Kbps and that MeeMix cannot perform volume leveling on the YouTube videos.

MeeMix is trying to keep users engaged while circumventing licensing issues, royalty fees, and bandwidth costs. Since all the videos are streamed from YouTube, it is Google’s problem if any copyright issues arise. (This is a similar approach taken by music search engine Songza, which streams the audio from YouTube videos to create music playlists).

Will the video channels help MeeMix attract more users, or is it too much a a MeeToo service?

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

Visual search and image recognition is one of the holy grails of consumer Internet technologies. Picitup is jumping into the deep end of this space by announcing the launch of its public beta.

Unlike Like.com (formerly Riya) which focuses on likeness, Picitup focuses on attaining matching images. This differentiation is important as it sets the company on a completely different trajectory in terms of both offering quantifiable value to users, as well as delivering a business model at the end of the day.

An image search on Picitup beings with a textual search actually queried on Google or Yahoo. Picitup will display a set of results only from one of the two—the basis of the decision is the speed and quality of the results. The user can then select which image Picitup should fetch similar images for, or filter the results by Faces, Products, Landscapes and Color. The analysis is made in real time and is based on 100+ parameters including a propriety color space the company developed.

Erick recently wrote that:

It’s hard to compete in the search engine market, but one approach taken by several startups is to sit on top of the big search engines and try to improve their results or interface. Why reinvent the wheel when you can simply add new spokes?

From a practicality point-of-view, relying on the likes of Google and Yahoo makes sense, but it should be noted that they forbid the reordering of their results, a sticking point that surely has a negative effect on the quality of results Picitup ultimately delivers.

Picitup claims it shortens the number of pages needed for an image search from 10 to 2. However, from my experimentation its engine’s match reliability was shaky. Results were pretty good for Ford Focus, but not even close for this Running Shoe. Note that all images should theoretically correspond to the top left-hand image.

Another issue that left a sour taste in my mouth was CelebrityMatchUp, an attempt to add some light-hearted fun to the beta interface. The idea here is that users upload photos of individuals and have Picitup produce results of people they resemble. This doesn’t exactly work. For instance, consider that Michael Arrington’s photo brought back results that he resembles both Barack Obama and John McCain. Huh?

Erick Schonfeld’s photo results are also somewhat curious, although the bright side is that Erick’s wife should be delighted to know she married a Kevin Costner look-alike. For a company claiming its forté is in image matching, Picitup should not have opened this door.

Alon Atsmon, co-founder & CEO, believes the company’s technology is compelling enough to drive revenue both from ads and through licensing a white label version of the engine for integration into ecommerce sites.

True, my initial impressions of Picitup are not necessarily positive ones. However, considering Atsmon is a serial entrepreneur I’ll remain optimistic and wait for Picitup to iterate a couple of more times before I cement my judgment.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

Many of us think we know what’s going to happen to the stock market. Few of us actually do. MarketGuru enables the more talented investors to reap the benefits of their predictive capacities and help others along the way.

While many social investment sites rank members almost entirely on their portfolio yields, the MarketGuru experience centers on trust. Your goal is to establish yourself as a “guru” - a member that other investors will turn to for advice. Gurus post their trade activity and motivations, which are then relayed to their followers through email. These followers can pay a premium of $30 per month to have messages from a maximum of three gurus delivered to them so they can execute their own transactions under similar market conditions.

The site currently offers no options for actual trading - instead, everything is carried out in a self-reported portfolio. MarketGuru hopes to implement real trading at some point in the future but it is still a long ways off.

Gurus are paid $5 per month for each follower they have, providing an incentive for them to attract and take care of their supporters. This meager bonus seems pretty lousy when compared to the vast sums of money that stand to be lost or gained on the market. I suspect that a few all-star gurus who collect hundreds of followers will do quite well for themselves, but most people will be left with just some additional pocket change.

MarketGuru’s concept has promise but the site is entering a space with a number competitors who provide very similar offerings. These include Covestor, CakeFinancial, SocialPicks, and Mint, which has just introducing investment tracking. That said, the social investment space can certainly support more than a few competitors, so MarketGuru has a fighting chance (especially with people who don’t actually want to play with real money).

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.

The ease of comparing prices on the Internet has done a lot to do away with major price differences between individual items at retailers, whether online or off. But where they still get you is when you buy many items from the same store and you throw in the high-margin coffee with the cut-rate shampoo. Higher-priced single items, such as a digital camera or an MP3 player, lend themselves more to online research. Finding the best deal is just a matter of selecting your preferred comparison shopping site (Shopping.com, mySimon, etc.). But what happens when you want to compare an entire cart of groceries across several merchants? Put simply, you are out of luck. Unless of course you happen to be living in the UK and making good use of mySupermarket.

mySupermarket, which has been around since 2006, claims to be the first comparison service that allows users to compare a cart of multiple items across retailers—in its case, groceries, across British supermarket chains Tesco, Sainsbury’s Ocado & ASDA.

It can compare not only identical items (a one-liter bottle of Coke), but also similar non-identical items, such one-liter bottles of mineral water from two separate brands. To accomplish this, mySupermarket classified 100,000 grocery products sold online in the UK according to multiple criteria and sub-criteria. The rules, weightings and relationships between different sub-criteria are incorporated into the company’s algorithms.

Up-to-date pricing is achieved using a combination of proprietary crawlers and manual validation processes to access real time prices from the supermarkets’ own sites. MySupermarket marries those prices with its own image database. It obtains products from the manufacturers and retailers and then uses in-house image production combined with post-processing facilities in Thailand.

The company claims an average online grocery cart includes approximately 50 items, with a total cost of 80-110 English pounds ($160-$220). By finding savings for consumers that average 20 percent per cart, and consumers accepting about half those recommendations, the actual savings average around 10 pounds ($20) per cart.

The way mySupermarket works is that users login to mySupermarket and fill-up a “trolley” (British for “shopping cart”). They are then presented with three types of recommendations:

1. Potential savings from switching the entire cart to another supermarket.
2. Potential savings from swapping items in the cart to alternatives from within the same supermarket.
3. Health conscious recommendations (calories, saturates, fat, salt, sugar) for swapping items to healthier alternatives.

The final step—payment—is actually performed on the desired supermarket’s own payment page. It should be noted that the service is absolutely free to consumers.

So where does mySupermarket derive its revenue from? Two sources: The first, targeted advertising based on the cart’s contents.

The second, a data service provided to the retailers and merchants which includes price listings, inventory listings (by zipcode), as well as comparison and analysis of products sold within the UK grocery sector. These days mySupermarket is focusing on expanding its UK business, as well as adding features, ad/promotion services and data reporting capabilities.

They are also considering requests to license their technology for non-grocery multi-item comparison shopping.

One of the key ingredients to success on the Web is rapid iteration, and to do so, eliciting user feedback is a must. As we know, TechCrunch is a breeding ground for avid beta testers keen to provide input and suggestions. Yet, more often than not, when we do offer feedback to a site all we receive in return is an auto-reply, thank-you email. When was the last time you submitted feedback to a site that was then followed up with an actual acknowledgment that the bug was fixed or the feature integrated? My guess, not very often.Kampyle—yet another Yossi Vardi startup from Israel—has developed a feedback management platform aimed at assisting site owners better manage this feedback loop and, along the way, increase customer loyalty and satisfaction. The underlying premise here being that users expect not only to be heard, but also responded to. This is especially true when providing feedback on services, products or customer experiences. Sites that manage their feedback right end up with deeper customer engagement, lower shopping cart abandonment, and better usability.

There are four moving parts in Kampyle’s feedback analysis platform:

1. Collection: A feedback collection form (see screenshot on right) is launched through buttons scattered across a site. Kampyle will roll out advanced customization options throughout the coming weeks.

2. Analysis: Kampyle provides user feedback along with contextual data such as screen resolution, browser type, operating system, etc. Slicing and grouping functionality delivers a greater understanding of why an issue occurred, which ultimately translates into what to do about it.
3. Management: Data and suggested corrective actions are arranged in intuitive dashboards. Data is exportable to XML & Excel. Nothing much to write home about here.

4. Action: CRM-like functionality completes the feedback loop by allowing site owners to inform groups and individual users of the corrective action they have taken. The notification even includes a link to where the fix was made and offers a thumbs-up/down rating to further express satisfaction.

All-in-all a wide variety of companies can benefit from a service such as Kampyle’s, which is bridging the gap between CRM and site-side analytics. Extending the CRM functionality to piggyback on full-blown CRM platforms such as SalesForce, SugarCRM, and even Zoho would go a long way in popularizing the service.

Kamplye is still in closed beta so pricing is yet to be established, but once the service is commercially available 250 TechCrunch readers will get their first month free. Sign-up here to get on the list.

This post was originally posted on TechCrunch.com where I cover the Israeli startup scene.