If search results that hit Yelp data linked to Yelp there'd be no problem. However if it links to Google Places because Google scraped the Yelp data and loaded it into the Places database - that's a problem.
Suppose someone launched an 'Encyclopedia Search' service that searched Wikipedia, Encyclopedia Britanica, etc and linked to their sites. The the service then started loading the Wikipedia and Britanica content into it's own respository and linked searches to that. It's still searching the same data, and taking you to the same articles, it's just that now you're not going to Wikipedia or Britanica anymore. Would that be ok? If not, what's the difference?
Poor analogy. Yelp contains facts and user reviews. Encyclopedias contain facts conveyed through written articles. The writing and reviews may be copyright protected. Facts are not.
Curated collections of facts are absolutely protected by copyright, provided they meet certain criteria. For example you can use that data to create a new curated data set for a different purpose, but you can't just copy all the data and use it in the way it was orriginaly compiled for because that's copying the curation. I can't say authoritatively if Yelp's data is likely to meet these criteria, but we're not talking about copyright violation anyway, we're talking about anti-competitive practices.
Using a dominance in search to scrape data from competitors for use in other services and then link to them from search in preference over the data's source may well be anti-competitive, even if the scraping itself isn't illegal.
Suppose someone launched an 'Encyclopedia Search' service that searched Wikipedia, Encyclopedia Britanica, etc and linked to their sites. The the service then started loading the Wikipedia and Britanica content into it's own respository and linked searches to that. It's still searching the same data, and taking you to the same articles, it's just that now you're not going to Wikipedia or Britanica anymore. Would that be ok? If not, what's the difference?