Method: TextPreprocessing#read_patterns

Defined in:
lib/automated_metareview/text_preprocessing.rb

#read_patterns(filename, pos) ⇒ Object

  • Reads the patterns from the csv file containing them.

    * maxValue is the maximum value of the patterns found



160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
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
 
# File 'lib/automated_metareview/text_preprocessing.rb', line 160

def read_patterns(filename, pos)
  num = 1000 #some large number
  patterns = Array.new
  state = POSITIVE
  i = 0 #keeps track of the number of edges
  
  #setting the state for problem detection and suggestive patterns
  if(filename.include?("prob"))
      state = NEGATED
  elsif(filename.include?("suggest"))
      state = SUGGESTIVE
  end
    
  FasterCSV.foreach(filename) do |text|
    in_vertex = text[0][0..text[0].index("=")-1].strip
    out_vertex = text[0][text[0].index("=")+2..text[0].length].strip

    first_string_in_vertex = pos.get_readable(in_vertex.split(" ")[0]) #getting the first token in vertex to determine POS
    first_string_out_vertex = pos.get_readable(out_vertex.split(" ")[0]) #getting the first token in vertex to determine POS
      
     patterns[i] = Edge.new("noun", NOUN)
     #setting the invertex
     if(first_string_in_vertex.include?("/NN") or first_string_in_vertex.include?("/PRP") or first_string_in_vertex.include?("/IN") or first_string_in_vertex.include?("/EX") or first_string_in_vertex.include?("/WP"))
          patterns[i].in_vertex = Vertex.new(in_vertex, NOUN, i, state, nil, nil, first_string_in_vertex[first_string_in_vertex.index("/")+1..first_string_in_vertex.length])
     elsif(first_string_in_vertex.include?("/VB") or first_string_in_vertex.include?("MD"))
      patterns[i].in_vertex = Vertex.new(in_vertex, VERB, i, state, nil, nil, first_string_in_vertex[first_string_in_vertex.index("/")+1..first_string_in_vertex.length])
     elsif(first_string_in_vertex.include?("JJ"))
      patterns[i].in_vertex = Vertex.new(in_vertex, ADJ, i, state, nil, nil, first_string_in_vertex[first_string_in_vertex.index("/")+1..first_string_in_vertex.length])
     elsif(first_string_in_vertex.include?("/RB"))
      patterns[i].in_vertex = Vertex.new(in_vertex, ADV, i, state, nil, nil, first_string_in_vertex[first_string_in_vertex.index("/")+1..first_string_in_vertex.length]) 
     else #default to noun
      patterns[i].in_vertex = Vertex.new(in_vertex, NOUN, i, state, nil, nil, first_string_in_vertex[first_string_in_vertex.index("/")+1..first_string_in_vertex.length])
     end      
     
     #setting outvertex
     if(first_string_out_vertex.include?("/NN") or first_string_out_vertex.include?("/PRP") or first_string_out_vertex.include?("/IN") or first_string_out_vertex.include?("/EX") or first_string_out_vertex.include?("/WP"))
      patterns[i].out_vertex = Vertex.new(out_vertex, NOUN, i, state, nil, nil, first_string_out_vertex[first_string_out_vertex.index("/")+1..first_string_out_vertex.length])
     elsif(first_string_out_vertex.include?("/VB") or first_string_out_vertex.include?("MD"))
      patterns[i].out_vertex = Vertex.new(out_vertex, VERB, i, state, nil, nil, first_string_out_vertex[first_string_out_vertex.index("/")+1..first_string_out_vertex.length])
     elsif(first_string_out_vertex.include?("JJ"))
      patterns[i].out_vertex = Vertex.new(out_vertex, ADJ, i, state, nil, nil, first_string_out_vertex[first_string_out_vertex.index("/")+1..first_string_out_vertex.length-1]);
     elsif(first_string_out_vertex.include?("/RB"))
      patterns[i].out_vertex = Vertex.new(out_vertex, ADV, i, state, nil, nil, first_string_out_vertex[first_string_out_vertex.index("/")+1..first_string_out_vertex.length])  
    else #default is noun
      patterns[i].out_vertex = Vertex.new(out_vertex, NOUN, i, state, nil, nil, first_string_out_vertex[first_string_out_vertex.index("/")+1..first_string_out_vertex.length])
    end
    i+=1 #incrementing for each pattern 
  end #end of the FasterCSV.foreach loop
  num_patterns = i
  return patterns
end