Class: NER

Inherits:

Object

• Object
• NER

Defined in:

lib/rbbt/ner/NER.rb,
lib/rbbt/ner/rner.rb

Direct Known Subclasses

Abner, Banner, ChemicalTagger, NGramPrefixDictionary, OSCAR3, OSCAR4, RegExpNER, TokenTrieNER

Instance Method Summary

• #entities(text, protect = false, *args) ⇒ Object
• #extract(text) ⇒ Object
• #initialize(model = nil) ⇒ NER constructor

  A new instance of NER.

Constructor Details

#initialize(model = nil) ⇒ NER

# File 'lib/rbbt/ner/rner.rb', line 167

def initialize(model = nil)
  begin
    require 'CRFPP'
  rescue Exception
    require File.join(Rbbt.datadir, 'third_party/crf++/ruby/CRFPP')
  end

  model ||= File.join(Rbbt.datadir, + 'ner/model/BC2')

  @parser = NERFeatures.new(model + '.config')
  @reverse = @parser.reverse
  @tagger = CRFPP::Tagger.new("-m #{ model } -v 3 -n2")
end

Instance Method Details

#entities(text, protect = false, *args) ⇒ Object

# File 'lib/rbbt/ner/NER.rb', line 6

def entities(text, protect = false, *args)
  case
  when Array === text
    text.collect do |element|
      matches = entities(element, protect, *args)
      matches.each{|match|
        match.offset += element.offset if match.offset and element.offset
      }
      matches
    end.flatten
  when (Segmented === text and protect)
    entities(text.split_segments(true), protect, *args)
  else
    match(text, *args)
  end
end

#extract(text) ⇒ Object

# File 'lib/rbbt/ner/rner.rb', line 181

def extract(text)
  features = @parser.text_features(text)

  @tagger.clear
  features.each{|feats|
    @tagger.add(feats.join(" "))
  }

  @tagger.parse

  found = []
  mention = []

  @tagger.size.times{|i|
    label = @tagger.y(i)
    word  = @tagger.x(i,0)

    if word == ')' 
      mention.push(')') if mention.join =~ /\(/
      next
    end

    case label
    when 1
      if mention.any? && ( mention.join(" ").is_special? || mention.select{|m| m.is_special?}.any?)
        found.push(mention)
        mention = []
      end
      mention.push(word)
    when 2
      mention.push(word)
    when 0
      found.push(mention) if mention.any?
      mention = []
    end
  }

  found << mention if mention.any?

  found.collect{|list| 
    list = list.reverse if @reverse
    list.join(" ")
  }
end