Sha256: 940d7f4f907f8776f0ea26e0d3ec9c9deb64bcaebc208f26ddf6dcc6aeb225de
Contents?: true
Size: 937 Bytes
Versions: 2
Compression:
Stored size: 937 Bytes
Contents
module Semantic module Transform class TFIDF @@number_of_documents_with_term = [] def self.transform!(matrix) number_of_documents = matrix.size2 @@number_of_documents_with_term = [] matrix.transpose.enum_for(:each_row).with_index do |document, column_index| document.enum_for(:each).with_index do |term_weight, row_index| unless term_weight == 0.0 inverse_document_frequency = 1 + GSL::Sf.log(number_of_documents / (number_of_documents_with_term(row_index, matrix).to_f + 1)) term_frequency = Math.sqrt(term_weight) matrix[row_index, column_index] = term_frequency * inverse_document_frequency end end end end def self.number_of_documents_with_term(row_index, matrix) @@number_of_documents_with_term[row_index] ||= matrix.row(row_index).where.size end end end end
Version data entries
2 entries across 2 versions & 1 rubygems
Version | Path |
---|---|
rsemantic-0.2.1 | lib/semantic/transform/tf_idf_transform.rb |
rsemantic-0.2.0 | lib/semantic/transform/tf_idf_transform.rb |