How to optimize selection with simultaneous sorting?

Question

Algorithms

There are a large number of elements (in RAM), each of which has a certain set of tags and a weight (a number, a criterion for sorting). You must first select all the elements by 2-3 tags, and then select 10 items with the highest weight from the result.
With tags, everything seems to be simple - it's hashing. If you scatter references to elements across hash tables, then you can select all the necessary ones in almost O (1). But then, in any case, you have to sort. Is it possible to somehow optimize this sorting or get rid of it? Because every microsecond counts.

Answer 1

In general, it turned out that it is possible to significantly speed up the algorithm if the nature of the initial data is at least a little known.
Of course, you can always deliberately slip such data that the algorithm will not be effective. But opposition to this is beyond the scope of the question.

Answer 2

Heap . Build a pile 10 long as you choose. A heap can be represented very efficiently by an array, and has good implementations in almost all languages. Your task is an academic example of heap application.