Source Code for Module featuregenerator.diff_generator

 1  """ 
 2   
 3  @author: Eleftherios Avramidis 
 4  """ 
 5  from __future__ import division 
 6  from featuregenerator import FeatureGenerator 
 7   
 8   
 9 -class DiffGenerator(FeatureGenerator): 
10      """ 
11      Operates on a ParallelSentence with two target sentences (pairwise). Computes subtraction of numerical features with the same name in the two target sentences 
12      """ 
13   
14   
15 -    def get_features_parallelsentence(self, parallelsentence): 
16          """ 
17          Gets executed once per parallel sentence. Performs subtraction of the respective numerical features of the 2 target sentences. 
18          Features with the same name get subtracted and the new feature gets added to the level of the parallel sentence.  
19          This is because the feature generation is being used on the last part of the generation process, upon pairwise comparison 
20          @param parallesentence: the object of the parallelsentence, already containing the simplesentences for the target translations 
21          @type parallelsentence: sentence.parallelsentence.ParallelSentence   
22          """ 
23          translations = parallelsentence.get_translations() 
24          ps_attributes = {} 
25          if len(translations)!=2: #diff features make sense only for pairwise comparisons 
26              return ps_attributes 
27          tgt1_attributes = translations[0].get_attributes() 
28          tgt2_attributes = translations[1].get_attributes() 
29          for tgt1_attribute_key in tgt1_attributes: 
30              if tgt1_attribute_key in tgt2_attributes: 
31                  #Check if the attribute is actually a float/int number 
32                  try:  
33                      tgt1_value = float(tgt1_attributes[tgt1_attribute_key]) 
34                      tgt2_value = float(tgt2_attributes[tgt1_attribute_key]) 
35                  except ValueError: 
36                      #if not possible, jump to the next attribute 
37                      continue  
38                  #calculate difference 
39                  diff = tgt2_value - tgt1_value 
40                  att_name = "%s_diff" % tgt1_attribute_key 
41                  ps_attributes[att_name] = str(diff) 
42          return ps_attributes 
43