VentureBeat

Though still arguably in its juvenile stages, the online advertising market has already been through several major technological shifts, with the introduction of behavioral, contextual and demographic targeting for ads. What’s next has been a subject for debate. Several early pioneers have bet that it’s semantic technology, with a company called Peer39.

In stealth mode for the last two years, Peer39 has begun slowly unveiling its product. Like other companies based on semantics and natural language processing, Peer39 claims its software, SemanticMatch, can read content in a way that roughly parallels human understanding.

Where for an advertiser like Coca-Cola, Google would find keywords like “cola” to bring up its ads, and other schemes would look at the age or location of web users, semantics take a more holistic look at a web page, determining the overall subject and tone.

Where keyword advertising might insert an ad because “cola” is used several times on the page, even if the overall topic is unrelated or the writing it negative to cola, semantic technology claims to be able to tell if an ad is really appropriate to the content.

Because of that, Peer39 could be useful for content that changes often and unpredictably — some examples being blogs, forums, news, and especially social media, which I speculated was the company’s target last year. User-generated content has been a particular challenge for advertising networks, so if Peer39 can prove SemanticMatch is effective, it’ll have its work cut out for it. To help prove itself, the company also has a a set of analytical tools based on semantic mapping, which help correlate ad placement to conversion and click-through rates.

It’s likely we’ll see some fast development in this space over the next year or two. Some have been speculating that Microsoft intends to move into the space after we broke the story that it plans to acquire semantic search company Powerset yesterday. Google, also, has been toying with its algorithms, though it’s as tight-lipped as ever. And other companies, like Ad Pepper Media, also say they’re developing semantic ad targeting.

While there seems to be enough money sloshing around for the best contenders to all get a piece, there are enough changes in online advertising, as well as mobile, to keep everything uncertain. VentureBeat contributor Julie Ruvolo wrote a lengthy piece about the view from Madison Avenue earlier this year that’s worth a look.

Peer39’s firepower comes from a set of executives and board members with backgrounds in either semantics or advertising. CEO Amiad Solomon sold a company called IDX to GE, while COO Matthew Goldstein just moved over from advertising giant Tacoda. Another former Tacoda Exec, Daniel Jaye, is on the board, along with Eytan Elbaz, who helped invent Google AdSense and sold Applied Semantics to the search giant.

The New York-based company has taken almost $12 million to date, between an $8.2 million funding we reported in October and a round of over $3 million earlier in the year. Its backers include Canaan Partners, Dawntreader Ventures and former Shopping.com CEO Daniel Ciporin.

When we broke the news of peer39’s initial $3 million funding in March, we reported that the company was working on a platform to use natural language processing for advertising. Less than six months later, we’ve learned that the startup has taken another $8.2 million.

The company, although trying to remain secretive, tells us it will soon be ready to launch. Peer39 combines semantic technologies with NLP to produce “groundbreaking algorithms that boost ROI for the industry,” according to a spokesman.

Back in March, investor Ed Sim of Dawntreader Ventures said on his blog, in response to our article, that the company mines unstructured data to boost conversion rates. We’re assuming it will attempt to make sense of the meaning of content on web pages, thus allowing advertisers to target ads more efficiently.

Another ad platform that recently released a so-called “semantic” advertising product is the United Kingdom’s Ad Pepper Media, which claims its iSense product can analyse every word on a web page to identify the overall subject, and then deliver an appropriate ad.

Google can do nearly as much with AdSense, though, so we think there might be more to peer39’s plan — in other words, going beyond word meaning toward sentence and overall story meaning.

Another clue might be the new website Twine, which also works with NLP and semantic search. Founder Nova Spivack told us recently that the company could potentially leverage its technology to provide users with highly targeted ads.

Twine is in the business of information collection — so it would have a large amount of the user’s personal info to work against. If peer39 wanted to do the same thing, it would need to be placed on websites where users directly interact with data.

That could be anything, from search engines to email clients. But it’s social networks, in particular, that are in desperate need of this kind of technology if they want to reach high valuations.

Of course, we’re just speculating.

The $8.2 million round of funding was led by Warren Lee at Canaan Partners, who will join the board. JPMorgan also participated. The previous round of $3 million came from Dawntreader Ventures.

Greenplum, a San Mateo company offering businesses an affordable way to sort though hundreds of terabytes of data to become more intelligent about their customers’ habits, has raised $19 million in financing.

Greenplum is significant because it says it provides a database for speedy data warehousing at a tenth of the cost of leading incumbent, Teradata. It does so by working with the new server built by Sun co-founder Andy Bechtolsheim, billed internally as the “Web 2.0 server.”

Called the Sun Fire X4500, the server has 48 disk drives and 24TB of storage, a significant amount considering the server is the size of two telephone books. (See Bechtolsheim demo the server in YouTube video below). Tim O’Reilly, a publisher who has in recent years become obsessed with collecting and processing business data to make better business decisions, has used it (see here for more details and a video where O’Reilly speaks with the former Greenplum CEO). O’Reilly discusses how Amazon and other Internet companies will need this technology, because they want to measure clicks, and other customers actions — to understand trends and find ways to boost profits. Greenplum has 15 customers, and announces tomorrow (Tuesday) that it has signed up the largest Phillipines wireless service provider, SMART, which also serves the most SMS messages in the world (the Filipinos really like texting).

Sierra Ventures led a $15 million cash investment, and was joined by existing investors Mission Ventures, Dawntreader Ventures, and EDF Ventures. Comerica Bank provided $4 million in debt. Notably, Sierra was the investor in competitor Teradata. Greenplum sells 100 TB hardware and software for $1.8 million, a tenth of Teradata’s price, co-founder Luke Lonergan tells VentureBeat. Teradata had $1.5 billion in sales last year, and revenue growth of 7.5 percent.

The company named Sun executive Bill Cook as chief executive. Greenplum is built on the high-end open source database, PostgreSQL