Module: NLP

Extended by:
LocalPersist
Defined in:
lib/rbbt/nlp/nlp.rb,
lib/rbbt/nlp/genia/sentence_splitter.rb

Defined Under Namespace

Modules: GdepChunk, GdepToken

Constant Summary collapse

NEW_LINE_MASK = 
"\t\t \t  \t"

Class Method Summary collapse

Class Method Details

.event_extraction(text) ⇒ Object 



95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
 
# File 'lib/rbbt/nlp/genia/sentence_splitter.rb', line 95

def self.event_extraction(text)
  events = ""
  marks = ""

  eventCount = 0

  pat = / [^ ]+[.!\?\)\]\"]( +)[^ ]+ /
  for line in text.split(/\n/) do
    while line.match(pat) do
      line.sub!(/ ([^ ]+)([.!\?\)\]\"])( +)([^ ]+) /){
        a, b, d, c = $1, $2, $3, $4
        events << eventCount.to_s  << "\t"
        events << returnFeatures(a, b, c)
        (" " << a << b << "__" << eventCount.to_s << "____" << d << "__" << c << " ")
      }
      eventCount += 1
    end
    marks << line
  end

  [events, marks]
end

.gdep_chunk_sentences(sentences) ⇒ Object 



148
149
150
151
152
153
154
 
# File 'lib/rbbt/nlp/nlp.rb', line 148

def self.gdep_chunk_sentences(sentences)
  sentences = Array === sentences ? sentences : [sentences]
  NLP.gdep_parse_sentences_extension(sentences).zip(sentences).collect do |segment_list, sentence|
    chunk_list = NLP.gdep_chunks(sentence, segment_list)
    NLP.merge_vp_chunks(chunk_list)
  end
end

.gdep_chunks(sentence, segment_list) ⇒ Object 



56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
 
# File 'lib/rbbt/nlp/nlp.rb', line 56

def self.gdep_chunks(sentence, segment_list)
  chunks = []

  chunk_start = "B"[0]
  chunk_inside = "I"[0]

  last = GdepToken.setup("LW")

  chunk_segments = []
  segment_list.each do |segment|
    if segment.chunk[0] == chunk_inside and not segment.offset.nil?
      chunk_segments << segment
    else
      if chunk_segments.any?
        cstart = chunk_segments.first.offset
        cend = chunk_segments.last.end
        chunk = sentence[cstart..cend]
        GdepChunk.setup(chunk, cstart, last.chunk.sub(/^.-/,''), chunk_segments)
        chunks << chunk
      end

      if segment.offset.nil?
        chunk_segments = []
      else
        chunk_segments = [segment]
      end
    end
    last = segment
  end

  if chunk_segments.any?
    cstart = chunk_segments.first.offset
    cend = chunk_segments.last.end
    chunk = sentence[cstart..cend]
    GdepChunk.setup(chunk, cstart, last.chunk.sub(/^.-/,''), chunk_segments)
    chunks << chunk
  end

    
  chunks
end

.gdep_parse_sentences(sentences) ⇒ Object 



98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
 
# File 'lib/rbbt/nlp/nlp.rb', line 98

def self.gdep_parse_sentences(sentences)
  sentences = Array === sentences ? sentences : [sentences]

  input = sentences.collect{|sentence| sentence.gsub(/\n/, NEW_LINE_MASK)} * "\n"
  sentence_tokens = TmpFile.with_file(input) do |fin|
    out = local_persist(Digest::MD5.hexdigest(input), :Chunks, :string) do
      CMD.cmd("cd #{Rbbt.software.opt.Gdep.produce.find}; ./gdep #{ fin }").read
    end

    out.split(/^$/).collect do |sentence|
      tokens = sentence.split(/\n/).collect do |line|
        next if line.empty?
        num, token, lemma, chunk, pos, bio, link, dep = line.split(/\t/)
        GdepToken.setup(token, nil, num, lemma, chunk, pos, bio, link, dep)
      end.compact
    end
  end

  sentences.zip(sentence_tokens).collect do |sentence, tokens|
    Segment.align(sentence, tokens)
  end
end

.gdep_parse_sentences_extension(sentences) ⇒ Object 



122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
 
# File 'lib/rbbt/nlp/nlp.rb', line 122

def self.gdep_parse_sentences_extension(sentences)
  require Rbbt.software.opt.Gdep.produce.ruby["Gdep.so"].find
  gdep = Gdep.new
  if not gdep.gdep_is_loaded
    Misc.in_dir Rbbt.software.opt.Gdep.produce.find do
      gdep.load_gdep 
    end
  end

  sentences = Array === sentences ? sentences : [sentences]

  sentence_tokens = sentences.collect{|sentence|
    Gdep.new.tag(sentence).split(/\n/).collect do |line|
      next if line.empty?
      token, lemma, pos, chunk = line.split(/\t/)
      GdepToken.setup(token, nil, nil, lemma, chunk, pos)
      token
    end.compact
  }

  sentences.zip(sentence_tokens).collect do |sentence, tokens|
    Segment.align(sentence, tokens)
    tokens
  end
end

.geniass_sentence_splitter(text) ⇒ Object 



249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
 
# File 'lib/rbbt/nlp/genia/sentence_splitter.rb', line 249

def self.geniass_sentence_splitter(text)
  Rbbt.software.opt.Geniass.produce
  offsets = []

  cleaned = text.gsub("\n",NEW_LINE_MASK)
  TmpFile.with_file(cleaned) do |fin|
    TmpFile.with_file do |fout|
      CMD.cmd("cd #{Rbbt.software.opt.Geniass.find}; ./geniass #{ fin } #{ fout }")

      
      Open.write(fin, Open.read(fin).gsub(NEW_LINE_MASK, "\n"))
      Open.write(fout, Open.read(fout).gsub("\n", '|').gsub(NEW_LINE_MASK, "\n"))
      # Addapted from sentence2standOff.rb in Geniass package

      inTxtStrict = Open.open(fin)
      inTxtNew = Open.open(fout)

      marker = "|"[0]
      position = 0
      sentenceCount = 1
      target = ''
      targetNew = ''
      start = 0
      finish = 0

      while(!inTxtNew.eof?) do
        targetNew = inTxtNew.getc
        target = inTxtStrict.getc
        position += 1
        if targetNew == marker
          sentenceCount += 1
          finish = position - 1
          offsets << [start, finish] if finish - start > 10
          if targetNew == target
            start = position
          else
            targetNew = inTxtNew.getc
            while targetNew != target do
              target = inTxtStrict.getc
              position += 1
            end
            start = position - 1
          end
        end
      end

      finish = position - 1
      offsets << [start, finish] if finish > start

      inTxtStrict.close
      inTxtNew.close
    end
  end

  offsets.collect do |s,e|
    sentence = text[s..e]
    next if sentence.nil?
    sentence.gsub!(NEW_LINE_MASK, "\n")
    Segment.setup sentence, s
    sentence
  end
end

.geniass_sentence_splitter_extension(text) ⇒ Object 



175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
 
# File 'lib/rbbt/nlp/genia/sentence_splitter.rb', line 175

def self.geniass_sentence_splitter_extension(text)
  cleaned = text.gsub("\n",NEW_LINE_MASK)
  events, marks = event_extraction(cleaned)

  Rbbt.software.opt.Geniass.produce
  begin
    ENV["LD_LIBRARY_PATH"] = Rbbt.software.opt.Geniass.lib.find + ":" + ENV["LD_LIBRARY_PATH"]
    require Rbbt.software.opt.Geniass.ruby["Geniass.so"].find
  rescue LoadError
    raise LoadError, "Geniass ruby module needs to be able to find #{Rbbt.software.opt.Geniass.lib.find} in LD_LIBRARY_PATH"
  end

  geniass = Geniass.new
  if not geniass.geniass_is_loaded
    Misc.in_dir Rbbt.software.opt.Geniass.find do
      geniass.load_geniass
    end
  end

  labels = events.split(/\n/).collect{|line| 
    geniass.label(line)
  }

  out = process_labels(marks, labels)

  offsets = []

  inTxtStrict = StringIO.new text
  inTxtNew = StringIO.new out.gsub("\n", '|').gsub(NEW_LINE_MASK, "\n")

  marker = "|"[0]
  position = 0
  sentenceCount = 1
  target = ''
  targetNew = ''
  start = 0
  finish = 0

  while(!inTxtNew.eof?) do
    targetNew = inTxtNew.getc
    target = inTxtStrict.getc
    position += 1
    if targetNew == marker
      sentenceCount += 1
      finish = position - 1
      offsets << [start, finish] if finish - start > 10
      if targetNew == target
        start = position
      else
        targetNew = inTxtNew.getc
        while targetNew != target do
          target = inTxtStrict.getc
          position += 1
        end
        start = position - 1
      end
    end
  end

  finish = position - 1
  offsets << [start, finish] if finish > start

  inTxtStrict.close
  inTxtNew.close

  offsets.collect do |s,e|
    sentence = text[s..e]
    next if sentence.nil?
    Segment.setup sentence, s
    sentence
  end

end

.merge_vp_chunks(chunk_list) ⇒ Object 



35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
 
# File 'lib/rbbt/nlp/nlp.rb', line 35

def self.merge_vp_chunks(chunk_list)
  vp = nil
  new_chunks = []
  chunk_list.each do |chunk|
    if chunk.type =~ /^VP/
      if vp.nil?
        vp = chunk
      else
        vp << chunk
        vp.parts.concat chunk.parts
      end
    else
      new_chunks << vp if not vp.nil?
      new_chunks << chunk
      vp = nil
    end
  end

  new_chunks
end

.process_labels(marked_text, labels) ⇒ Object 



139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
 
# File 'lib/rbbt/nlp/genia/sentence_splitter.rb', line 139

def self.process_labels(marked_text, labels)
  return "" if marked_text.empty? || labels.empty?
  out = ""

  count = 0
  text_lines = marked_text.split(/\n/)
  line = text_lines.shift
  for label in labels
    pat = "__" + count.to_s + "__"
    until(line.match(pat)) do
      out << line
      line = text_lines.shift
    end
    splitted = label.chomp.to_i

    line.sub!(pat){
      if splitted == 1
        "__\n__"
      else
        "____"
      end
    }
    line.sub!(/__\n____ +__/, "\n")
    line.sub!(/______( +)__/){
      $1
    }
    count += 1
  end

  out << line

  out << text_lines * ""

  out
end

.returnFeatures(prevWord, delimiter, nextWord) ⇒ Object 



7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
 
# File 'lib/rbbt/nlp/genia/sentence_splitter.rb', line 7

def self.returnFeatures(prevWord, delimiter, nextWord)
  if nextWord.match(/__ss__/)
    nw = nextWord.sub(/__ss__/, "")
  else
    nw = nextWord
  end

  str = ""
  # prev. word, next word
  str += "pw_" + prevWord.downcase
  str += "\tnw_" + nw.downcase

  # delimiter
  str += "\td_" + delimiter

  # capitalized first char in next word
  # capital in next word excluding first char.
  if nw[0].chr == nw[0].chr.capitalize
    str += "\tnfc_y"
    nwExcluginFirst = nw[1 ... -1]
    if nwExcluginFirst == nil
      str += "\tnwcef_n"
    elsif nwExcluginFirst.downcase == nwExcluginFirst
      str += "\tnwcef_n"
    else
      str += "\tnwcef_y"
    end
  else
    if nw.downcase == nw
      str += "\tnwcef_n"
    else
      str += "\tnwcef_y"
    end
    str += "\tnfc_n"
  end

  # prev. word capital
  if prevWord.downcase == prevWord
    str += "\tpwc_n"
  else
    str += "\tpwc_y"
  end

  # number in prev. word, in next word
  if prevWord.match(/[0-9]/)
    str += "\tpwn_y"
  else
    str += "\tpwn_n"
  end
  if nw.match(/[0-9]/)
    str += "\tnwn_y"
  else
    str += "\tnwn_n"
  end

  # prev., next word excluding braket, camma, etc.
  prevWordEx = prevWord.gsub(/[()'",\[\]]/, "")
  nwEx = nw.gsub(/[()'",\[\]]/, "")
  str += "\tpwex_" + prevWordEx.downcase
  str += "\tnwex_" + nwEx.downcase

  # bracket or quatation in prev. word
  if prevWord.match(/()'"/)
    str += "\tpwcbq_y"
  else
    str += "\tpwcbq_n"
  end
  # camma in prev., next word
  if prevWord.match(/,/)
    str += "\tpwcc_y"
  else
    str += "\tpwcc_n"
  end
  if nw.match(/,/)
  else
    str += "\tnwcc_n"
  end

  # prev. word + delimiter
  str += "\tpw_" + prevWord + "_d_" + delimiter
  # prev. word ex. +  delimiter + next word ex.
  str += "\tpwex_" + prevWordEx + "_d_" + delimiter + "_nwex_" + nwEx
  #str +=
  #str +=
  #str +=
  str += "\n"
end