Sort:  

Each object type has its own set of rankings. For example, restaurants are rated based on Ambiance, Service, Food, and Value:

image.png

For movies, there could be its own set of rankings, such as Acting, Direction, Plot, etc.

Each vote is recorded on the blockchain and signed by the user. As a result, it is easy to trace the votes of people I follow (trust) and compile a list of recommendations for discovering movies to watch.
Alternatively, it is possible to find like-minded individuals if their votes tend to match your own, and you can compile a list of movie suggestions based on their votes for movies which you have not yet watched (ranked).
With each vote a user casts, the system will be able to refine suggestions even more effectively. We are talking not just about finding similar movies, but rather about identifying films that have been positively rated by like-minded individuals.

Thanks for the detailed response. That's your solid structured data layer ;)