Python movie_reviews.categories函数代码示例

    def prep_reviews_data(self): # messy code to test classifier with movie reviews
        if not self.movie_review_data:
            print 'Preparing movie reviews...\n'
            from nltk.corpus import movie_reviews
            docs = [movie_reviews.raw(fileid) 
                    for category in movie_reviews.categories() 
                    for fileid in movie_reviews.fileids(category)]

            process = lambda x: 1 if x == 'pos' else -1
            labels = [process(category)
                    for category in movie_reviews.categories() 
                    for fileid in movie_reviews.fileids(category)]

            docs, labels = double_shuffle(docs, labels)
            training, testing = divide_list_by_ratio(docs)
            self.train_labs, self.test_labs = divide_list_by_ratio(labels)

            train_vecs = self.vectorizer.fit_transform(training)
            test_vecs = self.vectorizer.transform(testing)

            if isinstance(self.model, naive_bayes.GaussianNB):
                train_vecs = train_vecs.toarray()
                test_vecs = test_vecs.toarray()

            self.train_vecs = train_vecs
            self.test_vecs = test_vecs

            self.movie_review_data = True
            self.news_market_data = False

def download_data_if_not_yet():
    """
    Download the data set, if the data set is not download.
    """
    try:
        # make sure that nltk can find the data
        if paddle.v2.dataset.common.DATA_HOME not in nltk.data.path:
            nltk.data.path.append(paddle.v2.dataset.common.DATA_HOME)
        movie_reviews.categories()
    except LookupError:
        print "Downloading movie_reviews data set, please wait....."
        nltk.download(
            'movie_reviews', download_dir=paddle.v2.dataset.common.DATA_HOME)
        print "Download data set success....."
        print "Path is " + nltk.data.find('corpora/movie_reviews').path

def createFeatureSet(numOfExamples):    
    
    documents = [(list(movie_reviews.words(fileid)), category)
             for category in movie_reviews.categories()
             for fileid in movie_reviews.fileids(category)[:numOfExamples]]
    with open('documents.txt', 'wb') as f:        
        pickle.dump(documents, f)

##    #read from file
##    with open('documents.txt', 'rb') as f:
##        documents = pickle.load(f)

    random.shuffle(documents)
    
    all_words = []
##    for w in movie_reviews.words():
##        all_words.append(w.lower())
    #write to file
##    with open('allwords.txt', 'wb') as f:
##        pickle.dump(all_words, f)

    #read from file
    with open('allwords.txt', 'rb') as f:
        all_words = pickle.load(f)

    freqDist = nltk.FreqDist(all_words)
    #print('freq dist')
    #print(freqDist.most_common(50))

    word_features = freqDist.most_common(3000)

    featuresets = [(find_features(rev, word_features), category) for (rev, category) in documents]
    return featuresets

def main():
	documents = [(list(movie_reviews.words(fileid)),category) for category in movie_reviews.categories() for fileid in movie_reviews.fileids(category)]
	random.shuffle(documents)
	featuresets = [(document_features8b(d), c) for (d,c) in documents]
	train_set, test_set = featuresets[100:], featuresets[:100]
	classifier = nltk.NaiveBayesClassifier.train(train_set)
	print nltk.classify.accuracy(classifier, test_set)

def create_dataset():
    '''Create dataset from movie reviews dataset'''
    documents = [(list(movie_reviews.words(fileid)), category)
                for category in movie_reviews.categories()
                for fileid in movie_reviews.fileids(category)]
    random.shuffle(documents)
    return documents

def create_sentiment():
    """
        Train sentiment model and save.

        Input type: None 
        Output: Model as pickle 
    """

    random.seed(1)

    test = [
        ("The dude presenting Unravel seems like one of the most genuine game developers Ive ever seen I really hope this game works out for him",'pos'),
        ("His hands are shaking Dude looks so stoked and scared at the same time",'pos'),
        ("Right I just felt like I was watching his dream come true It was nice The game looks very well done as well Good for him",'pos'),
        ("Seriously Unravel looks really good actually and honestly seeing him so happy about what hes made is contagious I want to see more of Unravel ",'pos'),
        ("He was so nervous shaking all over his voice quivering",'neg'),
        ("The game looked nice too very cute art style ",'pos'),
        ("You could tell he genuinely wanted to be there it looked like he was even shaking from the excitement  I hope it works out for them aswell",'pos'),
        ("However following that up with the weird PvZ thing was odd To say the least",'neg'),
        ("Haha The game did look nice though Im definitely going to keep an eye on it I enjoy supporting such hopeful developers",'pos'),
        ("Very personable This looks like a buy for me As a dev in a other sector I appreciate this passion",'pos'),
        ("I want to give him a cookie",'pos'),
        ("Im getting a copy Im gonna support my indie devs",'pos'),
        ("The twitch leak was accurate It was like a play by play you start speaking French then switch to English",'neg'),
        ("yep exactly what i was thinking lol its important to note that the twitch leak never had them saying it was Dishonored 2 but that they were honored to be here very different",'neg'),
        ("Honored  Im 100 sure that was intentional",'neg'),
        ("oh yea for sure but wasnt solid enough evidence imo to be like dishonored 2 confirmed just based off that",'neg'),
        ("The confirmation was who was talking not what they were talking about ",'neg'),
        ("How awkward is it for a pop singer to perform at a video game conference",'neg'),
        ("Oh god did they warn him that he will get zero reaction",'neg'),
        ("I really hope so",'pos'),
        ("Almost as bad as Aisha fucking up her dialogue constantly Shes doing alright though E3 is really becoming a mainstream media event Hollywood has nothing like this ComicCon is the only comparison and they dont dazzle it up like E3",'neg')
        ]


    # Grab review data
    reviews = [
        (list(movie_reviews.words(fileid)), category)
        for category in movie_reviews.categories()
        for fileid in movie_reviews.fileids(category)
        ]
    random.shuffle(reviews)

    # Divide into 10% train/test splits
    new_train, new_test = reviews[:1900], reviews[1900:]

    # Train the NB classifier on the train split
    cl = NaiveBayesClassifier(new_train)

    # Compute accuracy
    accuracy = cl.accuracy(test + new_test)
    print("Accuracy: {0}".format(accuracy))

    # Show 5 most informative features
    cl.show_informative_features(5)

    # Save model for use in creating social model sentiment
    with open('sentiment_clf_full.pkl', 'wb') as pk:
        pickle.dump(cl, pk)
    print 'done saving model'

def f2c(corpus,fileName):
    if corpus=='mr':
        from nltk.corpus import movie_reviews as mr
        return mr.categories(fileids = fileName)[0]    
    else:
        from nltk.corpus import reuters
        return reuters.categories(fileids = fileName)[0]    

def prepareSentimentClassifier():

	documents = [(list(movie_reviews.words(fileid)), category)
		for category in movie_reviews.categories()
		for fileid in movie_reviews.fileids(category)]

	random.shuffle(documents)

	all_words = []
	for w in movie_reviews.words():
	    all_words.append(w.lower())

	all_words = nltk.FreqDist(all_words)
	
	global word_featuresSent
	word_featuresSent = list(all_words.keys())[:3000]

	featuresets = [(findFeaturesSentiment(rev), category) for (rev, category) in documents]
	
	training_set = featuresets[:1900]
	testing_set = featuresets[1900:]

	sentimentClassifier = nltk.NaiveBayesClassifier.train(training_set)

	print("Classifier accuracy percent:",(nltk.classify.accuracy(sentimentClassifier, testing_set))*100)

	return sentimentClassifier

def documentClassification():

    from nltk.corpus import movie_reviews

    documents = [(list(movie_reviews.words(fileid)), category)
            for category in movie_reviews.categories()
            for fileid in movie_reviews.fileids(category)]

    random.shuffle(documents)

    all_words = nltk.FreqDist(w.lower() for w in movie_reviews.words())
    word_features = all_words.keys()[:2000]

    def document_features(document):
        document_words = set(document)
        features = {}
        for word in word_features:
            features['contains(%s)' % word] = (word in document_words)
        return features

    print document_features(movie_reviews.words('pos/cv957_8737.txt')) 

    featuresets = [(document_features(d), c) for (d,c) in documents]
    train_set, test_set = featuresets[100:], featuresets[:100]
    classifier = nltk.NaiveBayesClassifier.train(train_set)

    print nltk.classify.accuracy(classifier, test_set)
    classifier.show_most_informative_features(5)

def category_by_movie():
    from nltk.corpus import movie_reviews as mr
    from nltk import FreqDist
    from nltk import NaiveBayesClassifier
    from nltk import classify
    from nltk.corpus import names
    from nltk.classify import apply_features
    import random

    documents = [(list(mr.words(f)), c) for c in mr.categories() for f in
mr.fileids(c)]
    random.shuffle(documents)

    all_words = FreqDist(w.lower() for w in mr.words())
    word_features = all_words.keys()[:2000]

    def document_features(document):
        document_words = set(document)
        features = {}
        for word in word_features:
            features['contains(%s)' % word] = (word in document_words)
        return features

    #print document_features(mr.words('pos/cv957_8737.txt'))
    #print documents[0]

    features = [(document_features(d), c) for (d, c) in documents]
    train_set, test_set = features[100:], features[:100]
    classifier = NaiveBayesClassifier.train(train_set)
    print classify.accuracy(classifier, train_set)

def main():
    """
    Sample training using the movie reviews corpus (Pang, Lee).
    """

    #== load inputs
    documents = np.array([movie_reviews.raw(review_id) 
        for category in movie_reviews.categories() 
        for review_id in movie_reviews.fileids(category)])

    sentiment_scores = np.array([0 if category == 'neg' else 1 
        for category in movie_reviews.categories() 
        for review_id in movie_reviews.fileids(category)])

    #== select random indices
    n = documents.shape[0]
    indices = np.random.permutation(n)
    threshold = np.floor(n*0.8) # 80% training set / 20% test set
    train_idx, test_idx = indices[:threshold], indices[threshold:]

    #== select training and validation sets according to these indicies
    x_train, x_test = documents[:, train_idx], documents[:, test_idx]
    y_train, y_test = sentiment_scores[:, train_idx], sentiment_scores[:, test_idx]

    #== train the model
    print '===== Training the model...'
    sentiment = SentimentMachine(x_train.tolist(), y_train.tolist())
    w = sentiment.train(speed=0.001, stochastic=False)
    print '===== Model trained.'

    #== test efficiency of the model
    print '===== Testing the model...'
    # compute the MSE
    h = lambda a,b: sigmoid(np.dot(a,b))
    x = sentiment.compute_features_matrix(x_test.tolist())
    mse = cost(w, x, y_test, h)
    # compute the number of valid classifications
    n_test = y_test.shape[0]
    valid = 0
    for i in xrange(n_test):
        valid += 1 if sentiment.classify(x_test[i]) == y_test[i] else 0
    percent = 100.0 * valid / n_test
    # print results
    print ('== Number of well-classified documents: {0} / {1} ({2}%)'
        .format(valid, n_test, percent))
    print '== Cost value on the test set: %.4f' % mse

def data_run():
    # print "Preparing Data..."
    labels = movie_reviews.categories()
    labeled_words = [(l, movie_reviews.words(categories=[l])) for l in labels]
    high_info_words = set(Params.high_information_words(labeled_words))
    feat_det = lambda words: Params.bag_of_words_in_set(words, high_info_words)
    feats = Train.label_feat_from_corps(movie_reviews, feature_detector=feat_det)
    return Train.split_label_feats(feats)

def label_docs():
    docs = [(list(movie_reviews.words(fid)), cat)
            for cat in movie_reviews.categories()
            for fid in movie_reviews.fileids(cat)]
    random.seed(42)
    random.shuffle(docs)

    return docs

    def __init__(self):
        self.documents = [(list(movie_reviews.words(fileid)), category)
             for category in movie_reviews.categories()
             for fileid in movie_reviews.fileids(category)]
        random.shuffle(self.documents)

        all_words = nltk.FreqDist(w.lower() for w in movie_reviews.words())
        word_features = all_words.keys()[:2000]

def get_documents():
    """
    Retrieve shuffled movie reviews from the nltk
    """
    print("Retrieving Movie Reviews\n")

    reviews = [(list(movie_reviews.words(fileid)), category)
               for category in movie_reviews.categories()
               for fileid in movie_reviews.fileids(category)]
    # so I have no idea why, but shuffle() gives me a none type
    return random.sample(reviews, len(reviews))