I intuitively believe that this will be an incredibly successful use case for Steem, but I'd like to make a number of points about how you have framed the idea.

1. Why would this be on Steemit.com rather than a completely different application? Steem is supposedly a platform, but Steemit.com is just a website. It shouldn't try to be everything to all people. Facebook is not Yelp, TripAdvisor and Google Places. Facebook has a very simple utility, sharing information via a post feed. Restaurant reviews is not it's forte. We need a new web app, like Steemap.com or SteemPlaces.com.

2. Why would you charge any of the businesses? Steemit.com doesn't charge anyone. We need the platform Steem to pay the apps in order for it to attract the developers and startups, just like it pays the contributors

3. We don't need any formal system for policing reviews. The system would do that itself. Falsifying claims would naturally be rewarded of confirmed by others. But only commonly available information could be rewarded. A claim that is not easy to disprove cannot be rewarded because rewards are based upon the weighted number of upvotes. So we're still only talking about the wisdom of the crowd, not really deep qualified reviews, unless there was some way of determining how qualified individuals were to provide their reviews. I don't know exactly how "miners" would verify that a review is legit or not. But a qualified restaurant reviewer who has shown himself/herself to be experienced in the trade would have a lot of weight on the platform.