Question

In a Ruby script,

• I have a hash that has sentences as keys and relevance scores as values.
• I want to retrieve an array containing the N most relevant sentences (highest scores).
• I want to retain the order in which these sentences are extracted.

Given:

hash = {
  'This is the first sentence.' => 5,
  'This is the second sentence.' => 1,
  'This is the last sentence.' => 6
}

Then:

choose_best(hash, 2)

Should return:

['This is the first sentence.', 'This is the last sentence.']

All the methods I can think of involve reordering the hash, thus losing the order of the sentences. What would be the best way to tackle this?

Answer 1

Try the following monster:

hash.map(&:reverse).each_with_index
                   .sort_by(&:first).reverse
                   .take(2)
                   .sort_by(&:last)
                   .map { |(_,s),_| s }

Another functional one:

hash.to_a.values_at(*hash.values.each_with_index
                         .sort.reverse
                         .map(&:last)
                         .sort.take(2))
         .map(&:first)

Note however, that as an unordered data structure, a hash table is not really suitable for this use case (although the order is remembered in Ruby 1.9). You should use an array instead (the sorting code remains the same):

sentences = [
  ['This is the first sentence.',  5],
  ['This is the second sentence.', 1],
  ['This is the last sentence.',   6],
]