A recommendation (or sometimes a tip) is a type of post that provides a short suggestion to another person. Often this is prompted by someone posting a recommendation request however, this could be sent unsolicited.
Request in Timeline
If a user sees a recommendation request on Facebook, they can click a button to search for recommendations within the correct geographic area and select the place they would like to recommend. They can also type in some additional information about why they are recommending the place.
Request with Recommendation Responses
All recommendations appear as items on the original recommendation request map as well as comments on the post, with a link preview about the location and the user's recommendation text.
Yelp both prompts users to add tips (somewhat akin to short reviews or comments) of their favorite dishes, features, or ideas related to particular locations so that they're easily searchable or find-able by future visitors of those locations. Often these are useful to future new visitors and function like recommendations typically made by staff, wait staff, or others who may be more biased.
Swarm / Foursquare
In an indieweb context this sort of functionality is likely to be achieved by accepting a Webmention on a post.
Eddie Hinkle - in thinking about this, I think this could be VERY cool when integrated into a reader. If there was parsable mf2 for a recommendation request. That way, if a reader detects a recommendation request, it could provide a similar interface to Facebook, allowing the user to search for a POI to recommend, and then it could create the post via Micropub, which would then be sent via webmention.
This could be very flexible as some readers could have built in recommendation posting UI (for example, that is likely something that Indigenous would do in the future). However, other readers like Monocle tend to push in-depth replies to Quill. This would work fine, because all the reader would have to do is pass the original url of the recommendation request to the other Micropub app (like Quill) and then the Micropub app can parse the url as it does for any reply functionality and it would get the geographic information from the recommendation request itself. This provides flexibility where readers can either have the micropub response functionality baked in or it can transfer the user out to a Micropub app.