Source Code for Module TEES.ExampleBuilders.FeatureBuilders.WordNetFeatureBuilder

 1  import sys, os 
 2  sys.path.append(os.path.join(os.path.dirname(os.path.abspath(__file__)), "../..")) 
 3  #print os.path.join(os.path.dirname(os.path.abspath(__file__)), "../..") 
 4  from Core.IdSet import IdSet 
 5  import Utils.Libraries.PorterStemmer as PorterStemmer 
 6  import Core.ExampleUtils as ExampleUtils 
 7  from FeatureBuilder import FeatureBuilder 
 8   
 9 -class WordNetFeatureBuilder(FeatureBuilder): 
10 -    def __init__(self, featureSet=None): 
11          FeatureBuilder.__init__(self, featureSet) 
12          from nltk.corpus import wordnet 
13          self.wordnet = wordnet 
14          print >> sys.stderr, "Using WordNet via NLTK" 
15       
16 -    def pennPOSToWordNet(self, pos): 
17          if pos.startswith("JJ"): 
18              return "a" #wn.ADJ 
19          elif pos.startswith("NN"): 
20              return "n" #wn.NOUN 
21          elif pos.startswith("VB"): 
22              #print "JEP" 
23              #print "VERB", wn.VERB 
24              return "v" #wn.VERB 
25          elif pos.startswith("RB"): 
26              return "r" #wn.ADV 
27          else: 
28              return None 
29       
30 -    def getSynset(self, text, wordNetPos): 
31          if wordNetPos == None: 
32              return None 
33          synsets = self.wordnet.synsets(text, pos=wordNetPos) 
34          if len(synsets) > 0: 
35              return [synsets[0]] 
36          else: 
37              return None 
38           
39 -    def getHypernyms(self, synset): 
40          rv = [] 
41          rv.append("HYPER_"+synset.name) # add also the base level 
42          for hypernym in synset.hypernyms(): 
43              rv.append("HYPER_"+hypernym.name) 
44          return rv 
45       
46 -    def getTokenFeatures(self, tokenText, pennPos): 
47          #print tokenText, pennPos, "X", 
48          rv = [] 
49          if tokenText == None: 
50              return rv 
51          #print "A", 
52          wordNetPos = self.pennPOSToWordNet(pennPos) 
53          #print "B", 
54          synsets = self.getSynset(tokenText, wordNetPos) 
55          #print "C", 
56          if synsets != None: 
57              rv.extend(self.getHypernyms(synsets[0])) 
58              rv.append("LEX_" + synsets[0].lexname) 
59          #print "D" 
60          return rv 
61   
62  if __name__=="__main__": 
63      w = WordNetFeatureBuilder() 
64      print w.getTokenFeatures("cat", "NN") 
65