With a large benchmark, some kind of structuring of the queries is desirable—a good taxonomy must be useful in this respect. The taxonomy is also useful when dividing the task of proposing queries among multiple contributors.