getAdjectives.py

import math
import string
import nltk
from nltk.corpus import brown as sc
from collections import Counter
from nltk.corpus import movie_reviews
from nltk.corpus import wordnet as wn
import MPQALexicon

# NLTK's pos tagger kinda sucks sometimes.  Let's double check with Wordnet.
def isAdj(word):
    global tagger, mpqa_words
    #return tagger.tag([word])[0][1].startswith("JJ")
    #return len(wn.synsets(word, wn.ADJ)) > 0
    #return tagger.tag([word])[0][1].startswith("JJ") and len(wn.synsets(word, wn.ADJ)) > 0 and word in mpqa_words
    return word in mpqa_words

def genConj():
    conj = open('movieconj.txt', 'w')
    global tagger
    tagger = nltk.tag.perceptron.PerceptronTagger()
    num_lines = 0
    for review in sorted(movie_reviews.fileids()):     #For every review
        tokens = movie_reviews.words(fileids=[review])
        for i in range(0,len(tokens)-2):
            if isAdj(tokens[i]) and isAdj(tokens[i+2]):
                line = tokens[i]+ " " + tokens[i+2] + " " + tokens[i+1] + "\n"
                print line.strip()
                if (tokens[i+1] == "and"):
                    conj.write(line)
                    num_lines += 1
                elif (tokens[i+1] == "or"):
                    conj.write(line)
                    num_lines += 1
                elif (tokens[i+1] == "but"+ "\n"):
                    conj.write(line)
                    num_lines += 1
                elif (tokens[i+1] == "either-or"):
                    conj.write(line)
                    num_lines += 1
                elif (tokens[i+1] == "neither-nor"):
                    conj.write(line)
                    num_lines += 1
    print num_lines
    
def doBrown():
    f = open('movieconj.txt', 'w')
    list1 = []
    for word in sc.tagged_sents():
        for w in word:
            if w[1].startswith("JJ"):
                list1.append(w[0])
    counts = Counter(list1)
    d = dict(counts)

    for n in d:
        if( d[n] >= 15):
            f.write(n+" \n")
    f.close()

(mpqa_words, mpqa_labels) = MPQALexicon.load()
genConj()