Dish.fm has relaunched its food recommendation app, moving away from crowdsourced reviews into sentiment analysis technology that is built on top of existing review data.

TechCrunch reports that the new iOS app is built on top of content shared on established websites, such as Yelp and Foursquare, with the company analysing more than 3m existing reviews on these sites to build up ‘likes’ within the app. This then translates into food recommendation at individual restaurants, although the app is limited to New York and San Francisco at launch with more US locations coming soon.

Users can also access pictures, read reviews and share their own opinions of individual dishes.

“Not all dishes taste the same – even at the best restaurants. Dish.fm helps you discover the best dishes at any restaurant with confidence,” the company says in the App Store description. “In one quick glance, learn what is the best dish, drink or dessert to order based on the results of millions of reviews.”