Что означает эта ошибка?

---

у меня есть код
и у меня есть эта ошибка:
ValueError: пустой словарь:возможно, документ содержит только стоп-слова.



на этом пока остановлюсь words.txt:
а то как в если

Что я уже пробовал:

#Created on Sat May 07 21:12:41 2016


from sklearn import cross_validation
from sklearn.datasets import load_files
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import TfidfTransformer
from sklearn.naive_bayes import MultinomialNB
from nltk import wordpunct_tokenize, PorterStemmer, LancasterStemmer
import os
import nltk, os.path ,sklearn
#path = os.path.abspath(D:\\IR\\train)
#fi = load(path)
count_vect = CountVectorizer()    
tfidf_transformer = TfidfTransformer()    
clf= MultinomialNB()

data = load_files(container_path='D:\\train')
stopWordsFile = open('D:\\IR\\IR-Project\\StopWord.txt','r')
stopWords = stopWordsFile.read().split(' ')
porter = PorterStemmer()

#min_df=1

def train():    
    print ("training")
    for index in xrange(len(data.data)):
        tempData = data.data[index]
        tokens =  wordpunct_tokenize(tempData)
        #stopWordsRemoved = [unicode(token, 'utf-8') for token in tokens if token not in stopWords]
        stopWordsRemoved = [token.decode("windows-1252") for token in tokens if token not in stopWords]
        stems = [porter.stem(word) for word in stopWordsRemoved]
        data.data[index] = str(stems)
        
    
        
    trainSpars = count_vect.fit_transform(data.data)
    trainTfidf = tfidf_transformer.fit_transform(trainSpars)
    print (trainTfidf.shape)
    clf.fit(trainTfidf,data.target)

def evaluate(testSentence):  
    print ("evaluating")
    tokens =  wordpunct_tokenize(testSentence)
    stopWordsRemoved = [token.decode("windows-1252") for token in tokens if token not in stopWords]
    stems = [porter.stem(word) for word in stopWordsRemoved]
    st = ' '
    result = st.join(stems)
    result = [result]
    print (result)
    spars=count_vect.transform(result)
    tfidf=tfidf_transformer.transform(spars)
    print (tfidf.shape)
    
    predicted=clf.predict(tfidf)
    predict_ranking = clf.predict_proba(tfidf)
    print ("choosen class " ,predicted.item())
    print ("ranking probability ", predict_ranking)
    return predicted , clf , tfidf

      
    

train()
my_str = input("Please inter desired sentence")
print (my_str)

testSentence = my_str
predict , clf ,tfidf  = evaluate (testSentence)

<pre lang="Python">

---