There are so many mobile apps floating around these days for books, music, games and TV that it can be a bit daunting to find the ones that matter to you. A Seattle upstart wants to help sort through the clutter. It’s called Mynd, and it’s the latest project from Greg Martin, Sam Teplitsky and Kaveh Ghadianipour at Liquify Digital.

The idea, as the company explains it, is to combine machine learning with curated app recommendations so that the best apps for each individual surfaces more readily.

“What separates Mynd from other app discovery solutions on the market is network intelligence: the more the user interacts with the network, the more intelligent the network becomes,” the company says.

I’ve been playing around with the app over the past day, using it to discover new apps in the books, music and games arena. At this point, I’ve been scanning the apps of everyone in the system, though there is the ability to track my friends’ top apps as well. You can also track the “stream” of activity in Mynd to see which apps are getting recommended. (I, for example, just learned of RunPee, which tells users the best time to go to the bathroom during a movie without missing any of the action. Now, that’s American ingenuity at work, folks).

Mynd isn’t the only one looking to help mobile phone users discover new apps. Seattle’s AppStoreHQ also offers a recommendation engine for mobile apps, most recently expanding the offering to include apps on Windows Phone.

At this point, Mynd is only available for iPhone apps. But the company has plans to expand onto other platforms.

The company gets affiliate revenue when a user follows a recommendation and downloads a paid app.

A new version of Mynd is slated for next week, with Martin saying that it’s designed to offer recommendations across categories. For example, if you like a certain game, you might like a specific book.

“We think by combining three human curation tiers with machine learning that pays attention to both consumer behavior and the emotional value points attached to each piece of content the user interacts with, we can make highly personalized recommendations which may have previously gone undiscovered,” says Martin, who previously worked at Scout Media in Seattle.

Here’s a more technical explanation of how Mynd works from Martin:

“Similar to how the Pandora Music Genome uses a database of over 400 attributes to apply to each piece of music, we will start with a set of 20 “Emotags” – 10 in the positive column and 10 in the negative column – and apply just three Emotags to each curated entertainment property … as it’s entered into the system. By combining these tags with the IP metadata and the machine learning from our user’s behavior, we believe we have found the bridge that will enable us to deliver accurate, personalized entertainment recommendations to users which crosses genres (e.g.; Horror to Drama in the case of Netflix) and even entire categories (e.g.; Games to Books in the case of Amazon).”

Comments

Job Listings on GeekWork