Jennifer Zaino
SemanticWeb.com Contributor

Former hedge fund manager and current CNBC TV host Jim Kramer has "boo-yah" as his catch phrase. Nagaraju Bandaru has "BooRah."

That's the name of the web site co-founded by Bandaru, also CTO of the semantic web-enabled personalized restaurant review site. ("Boo" for the bad restaurants, "Rah" for the good ones -- "sentiment in semantics," as Bandaru phrases it.) Started about three years ago, the private, venture-funded company has had a busy few weeks. In the last couple of weeks, it has announced a partnership with InfoUSA, one of the top providers of local business listings to every online directory, navigation and mobile player, to provide ratings and reviews for approximately 150,000 of InfoUSA's restaurant listings; an API for its local syndication partners with enhanced mobile platform capabilities; and a version of its application for the Google Android software stack that has had over 30,000 downloads in two weeks.

The inspiration for BooRah, Bandaru says, was the fact that there wasn't an easy way for people to get the information they needed out of the tons of formal and blog reviews of restaurants out there in their locales, and that tags didn't adequately communicate what people were trying to say. The site has these core semantic technology aspects:

The ability to match web page to the correct local business. Particularly in blogs, people might discuss restaurants very casually -- e.g., they had a great dinner at Joe's Pizza, Bandaru says. But which Joe's Pizza? "We are able to map any entity on any web page we crawl and associate that with the correct local business," he says.

The ability to extract attributes from web pages in a very scalable, automated way. "Each sentence is analyzed for any term and grouped into a top level category -- food, service or ambience," he says, and those terms contribute to overall ratings on each of those counts.

"At the macro level the real problem is you go to any review site you can find a rating, 3 or 3.5 or 4 stars. But it's never beyond those because that's how people write. But when they consume reviews they want more information -- is that 3 stars for food or ambience, so we can break down that rating into a percentage score of 100 and give a more granular score," with the help of its patented natural language processing capabilities. Since its launch, BooRah has grown from a vocabulary of about 1,000 to 50,000 terms.

The core technology in the semantic area is giving a quality score associated with sentiment. For example, if you say a restaurant was not so good vs. very bad, the technology would know that the former description rates slightly higher than the latter. "So the core of user reviews is all about negation, how do you handle people's slang, associate what they are describing and what category and roll it up across every single sentence in a review across every different source and summarize," Bandaru says.

Currently the site mines information from close to 100 formal review sites and 100,000 blogs. That adds up to over 1.5 million online reviews in the top 20 U.S. metro areas.

One of the reasons Bandaru sees opportunity for its site on mobile platforms is the shelf life of ratings-oriented applications. Most mobile applications are more gadgety in nature, he says, useful for a couple of days until people tire of them.

"What makes an application survive in a mobile situation for a long time is, what is the data behind the scenes that supports a real life application, and how effectively can you provide that in a contextually relevant fashion, or some other way of optimizing their search," he says. Mobile monetization is starting to become a reality for applications that can deliver that kind of value, he says. "Semantics is a pretty good angle, and there is a lot of contributing value to the future business model from optimizing ads, optimizing how people do searches and ultimately driving engagement, which is where people look to semantics to help right now."

BooRah's relationship with InfoUSA is one of three fairly big partnerships, the other two not yet announced. The InfoUSA partnership means that its customers can get BooRah's ratings and other semantically-enabled information as simple feeds to their own online sites. Its new Web service API that supports up to 5,000 queries a day is a step at supporting the development of location-enabled restaurant applications on mobile platforms, integrating with Mozilla Geode and Skyhook Loki.

Currently, between BooRah and its partner web sites, Bandaru counts 1 million visitors a month, and, he says, "we are definitely looking to expand beyond restaurants. That is the next step for us."