Ron Franczyk came up with the idea for his latest company  -- a Seattle semantic search startup called Sortuv -- after trying to find a new rug. Franczyk, who along with founders Andrew Newman and Jim Heising sold Giant Company Software to Microsoft in 2004, wanted a rug that "sort of" went with the new furniture he had just bought for his home. The only problem was, he couldn't quite describe what he was looking for when he went to the traditional search engines.

From that experience, Sortuv was born. The 14-month-old Pioneer Square startup is now emerging from stealth mode with a new search and discovery technology that helps people find information when keywords don't do the trick.

"The problem that we are faced with is that the possibilities are overwhelming when it comes to certain categories of things and really how do you find what you are looking for? And the more options that are available in the category, the more difficult the problem is," said Franczyk.

Sortuv tries to sort through that information by helping people find something that relates to something else. For example, it could help people find a bar or night club in downtown Seattle that is "sort of" like El Gaucho. Though it sounds odd, the technology could go so far as to find a bottle of wine that is like a Monet painting.

The key is what Heising calls a "huge node graph" that shows how various things are related to one another, similar to the recommendation engine that Amazon.com pioneered.

For example, it could find that two restaurants advertise in the same magazine or appeared in the same blog post.

Semantic search -- the idea of utilizing natural language to help people find more relevant information online -- is certainly a hot area filled with companies such as Powerset (acquired by Microsoft), Evri and Radar Networks (both bankrolled by Paul Allen).

But while Sortuv sometimes gets tossed in that bucket, Heising said it is different. That's because other semantic search startups that are attempting to figure out the meaning of sentences. Sortuv, on the other hand, is utilizing a statistical model to figure out connections.

"We don't necessarily need to know the meaning," said Heising. "We just need to know that these things were seen in proximity to one another, and over a certain number of times and over a certain number of connections it starts to become meaningful."

At this point, Sortuv is showcasing its technology with a Facebook and iPhone application called Near + Now that attempts to recommend restaurants, bars, concerts and other activities based on a user's specific tastes. It shows the places that are "most like you," said Heising. And it does that whether the person is in New York, Chicago or Seattle, finding connections between places and offering suggestions.

"You could even say: 'Find me a place like me in New York or find me a place like 13 Coins in New York City?" explains Heising. The service also incorporates past behaviors in order to make stronger recommendations, though Heising said that is not necessary for it to work.

Over time, the self-funded company plans to roll out a full-blown Web site and expand into new markets.

For example, Heising said the technology could be applied to human resources in order to help companies find job candidates who resemble other successful employees. Or, he said, it could be used to find patterns in legal documents. 

Given the dominance of music recommendation services such as Pandora, iLike and Last.fm, they've steered clear of the music business for now. But that's also a possibility down the road, he said.

When the team formed last year, they started kicking around the idea of building a new search engine.  But Heising said they quickly realized that that would be a "long and arduous process."

So, they started to apply the technology in the much narrower niche of local search, focusing on helping people find restaurants, bars, hotels and other businesses. Of course, that too is a crowded area with the likes of Yelp, CitySearch, Urbanspoon, GoTime and others.

But the team, which worked together at Giant Company and Click Commerce, thinks they have an unique approach by offering more meaningful recommendations. Even so, the company's ultimate goal is not to become a local search provider. Instead, it hopes to license the technology to companies in order to improve their search and discovery process.

For example, an iPhone application like Urbanspoon that offers restaurant recommendations could incorporate the technology to show restaurants that have more meaning to the user. That means if you gave the iPhone a shake, there might be a better chance you'd actually want to go to the restaurant.