『マルチメディアデータ工学:音声・動画像データベースの高速検索技術 (宝珍 輝尚)』31㌻のシグネチャファイルのRubyでの実装

signature_file.rb

# frozen_string_literal: true

KEYWORDS =
  [
    %w[ribbon trouble sister vigorous],
    %w[representative deny hypothesize]
  ]
  .freeze

QUERY_KEYWORD = 'representative'

to_signature = ->(text) { text.unpack1('b*').rjust(100, '0') }
binominal_disjunction = ->(args) { args.map(&:to_i).reduce(:|) }

greater_bits =
  lambda do |bits1, bits2|
    bits1
      .chars
      .zip(bits2.chars)
      .all? { |a, b| a == '1' || b == '0' }
  end

keyword_signatures =
  KEYWORDS.map do |keywords|
    keywords.map { |keyword| to_signature.call keyword }
  end

text_signatures =
  keyword_signatures.map do |signatures|
    signatures.first.chars.zip(*signatures.drop(1).map(&:chars))
              .map(&binominal_disjunction)
              .join
  end

query_signature =
  to_signature.call QUERY_KEYWORD

result =
  text_signatures.filter { |sign| greater_bits.call(sign, query_signature) }

puts result