Sentiment analysis using python

Question

I have some text files containing movie reviews I need to find out whether the review is good or bad. I tried the following code but it's not working:

import nltk
with open("c:/users/user/desktop/datascience/moviesr/movies-1-32.txt", 'r') as m11:
    mov_rev = m11.read()
mov_review1=nltk.word_tokenize(mov_rev)
bon="crap aweful horrible terrible bad bland trite sucks unpleasant boring dull moronic dreadful disgusting distasteful flawed ordinary slow senseless unoriginal weak wacky uninteresting unpretentious "
bag_of_negative_words=nltk.word_tokenize(bon)
bop="Absorbing Big-Budget Brilliant Brutal Charismatic Charming Clever Comical Dazzling Dramatic Enjoyable Entertaining Excellent Exciting  Expensive Fascinating Fast-Moving First-Rate Funny Highly-Charged Hilarious Imaginative Insightful Inspirational Intriguing Juvenile Lasting Legendary Pleasant Powerful Ripping Riveting Romantic Sad  Satirical Sensitive  Sentimental Surprising Suspenseful Tender Thought Provoking Tragic Uplifting Uproarious"
bop.lower()
bag_of_positive_words=nltk.word_tokenize(bop)
vec=[]
for i in bag_of_negative_words:
    if i in mov_review1:
        vec.append(1)
    else:
        for w in bag_of_positive_words:
            if w in moview_review1:
                vec.append(5)

So I am trying to check whether the review contains a positive word or a negative word. If it contains a negative word then a value of 1 will be assigned to the vector vec else a value of 5 will be assigned. But the output I am getting is an empty vector.

Please help. Also, please suggest others way of solving this problem.

Answer 1

Try to search from the databases of official "bad words" that google publishes in this link Google's official list of bad words. Also, here is the link for the good words Not the official list of good words

For the code, I would do it like this:

textArray = file('dir_to_your_text','r').read().split()

#Bad words should be listed like this for the split function to work
# "*** ****** **** ****" the stars are for the cenzuration :P
badArray = file('dir_to_your_bad_word_file).read().split()
goodArray = file('dir_to_your_good_word_file).read().split()

# Then you use matching algorithm from difflib on good and bad word for every word in an array of words
import difflib

goodMachingCouter = 0;
badMacihngCouter = 0;


for iGood in range(0, len(goodArray)):
    for iWord in range(0, len(textArray)):
        goodMachingCounter += difflib.SequenceMatcher(None, goodArray[iGood], textArray[iWord]).ratio()
     
for iBad in range(0, len(badArray)):
    for iWord in range(0, len(textArray)):
        badMachingCounter += difflib.SequenceMatcher(None, badArray[ibad], textArray[iWgoodord]).ratio()

goodMachingCouter *= 100/(len(goodArray)*len(textArray))
badMacihngCouter *= 100/(len(badArray)*len(textArray))

print('Show the good measurment of the text in %: '+goodMachingCouter)
print('Show the bad measurment of the text in %: '+badMacihngCouter)
print('Show the hootnes of the text: ' + len(textArray)*goodMachingCounter)

The code will be slow but accurate :) I didn't run and test it please do it for me and post the correct code :) because I wanna test it too :)

Answer 2

The following link contains a list of positive and negative polarised emotions on the scale of [-5, 5]. Just try to count up the scores based on the word matches and you can get the overall movie review score.

AFINN

Answer 3

try

vec =[]

for word in bag_of_negative_words:
    if word in mov_review1:
        vec.append(1)

for word in bag_of_positive_words:
    if word in moview_review1:
         vec.append(5)