If we would like to draw some connection between the map$-$reduce programming framework with the SQL queries that we have learned in classical databases, which of the following statements is the most accurate?

The map function is like going through each row to perform some operation, then we perform a GROUP BY between map and reduce, and the reduce function is like outputing an aggregate result for each group.

The map function is like the main part of a SQL query. Then there is some data shuffling between map and reduce. Finally, the reduce function corresponds to filters $($i$.$e$.,$ WHERE clause$)$ or projections or aggregates that can reduce the amount of data in the output.

The map function corresponds to applying predicates over each row of a database $($the cursor model$).$ Bwtween map and reduce there is some JOIN of data between relations. Then the reduce function aggregates the JOIN results and possibly performs projection and further filtering.

The map function is like creating a temporary view that contains all the needed data. There is some possible data shuffling between map and reduce due to the distributed nature of the computing environment. At last, the reduce function performs the final data reduction from the view and produces the output.