Graphing Live Twitter Sentiment 

The final stage of this set of tutorials is graphing the sentiment output…based on a more in depth tutorial here:-

https://pythonprogramming.net/live-graphs-matplotlib-tutorial/

Code here:-

import matplotlib.pyplot as plt
import matplotlib.animation as animation
from matplotlib import style
import time

style.use("ggplot")

fig = plt.figure()
ax1 = fig.add_subplot(1,1,1)

def animate(i):
    pullData = open("twitter-out.txt","r").read()
    lines = pullData.split('\n')

    xar = []
    yar = []

    x = 0
    y = 0

    for l in lines[-200:]:
        x += 1
        if "pos" in l:
            y += 1
        elif "neg" in l:
            y -= 1

        xar.append(x)
        yar.append(y)
        
    ax1.clear()
    ax1.plot(xar,yar)
ani = animation.FuncAnimation(fig, animate, interval=1000)
plt.show()

As I’m running this on a headless server it ran into issues straight away…

galiquis@localhost: $ python3 nltk_tutorial21.py
Unable to init server: Could not connect: Connection refused
Unable to init server: Could not connect: Connection refused

(nltk_tutorial21.py:26565): Gdk-CRITICAL **: 08:21:48.134: gdk_cursor_new_for_display: assertion ‘GDK_IS_DISPLAY (display)’ failed

(nltk_tutorial21.py:26565): Gdk-CRITICAL **: 08:21:48.137: gdk_cursor_new_for_display: assertion ‘GDK_IS_DISPLAY (display)’ failed

So after a little research I found a way of switching the ‘canvas’ to the Agg Buffer – allowing the output to be saved rather than shown.

import matplotlib as mpl
mpl.use('Agg')

It’s important that this is defined ahead of any other canvas calls/functions – otherwise it throws errors.

The other tweaks I made just switched off the animation for now.

import matplotlib as mpl
mpl.use('Agg')

import matplotlib.pyplot as plt
import matplotlib.animation as animation
#from matplotlib import style

import time

#style.use("ggplot")


fig = plt.figure()
ax1 = fig.add_subplot(1,1,1)

def animate(i):
    pullData = open("twitter-out.txt","r").read()
    lines = pullData.split('\n')

    xar = []
    yar = []

    x = 0
    y = 0

    for l in lines[-200:]:
        x += 1
        if "pos" in l:
            y += 1
        elif "neg" in l:
            y -= 1

        xar.append(x)
        yar.append(y)

    ax1.clear()
    ax1.plot(xar,yar)

#ani = animation.FuncAnimation(fig, animate, interval=100)
animate(1)
fig.savefig('temp.png')

Twitter sentiment analysis

First up the Twitter API module needed installing:- 

galiquis@raspberrypi: $ pip3 install tweepy

Next a Twitter App is required from this link:-

https://developer.twitter.com/en/apps

This required setting up a developer account – with more justification in the application form than I was expecting – especially around what I’d be using the app for….anyway once generated it gave a live stream of twitter based on this code:-

from tweepy import Stream
from tweepy import OAuthHandler
from tweepy.streaming import StreamListener

#consumer key, consumer secret, access token, access secret.
ckey="6ru23AnzOKAieH4eYXF0XuTPS"
csecret="74Oz560aRCfo5QzzXu2I0gfOm58qkNPfZx0oSl3tnWEnEND4ex"
atoken="241873929-QkQ1eN0Du1Cg6el6rJa3sMGRHBaiSp7Cxekq61Of"
asecret="hf0ECMVfcqlPWkgOGKeNNTU1m41QQuiTOLzktsiNqqIxD"

class listener(StreamListener):

    def on_data(self, data):
        print(data)
        return(True)

    def on_error(self, status):
        print(status)

auth = OAuthHandler(ckey, csecret)
auth.set_access_token(atoken, asecret)

twitterStream = Stream(auth, listener())
twitterStream.filter(track=["car"])

https://pythonprogramming.net/twitter-api-streaming-tweets-python-tutorial/

The below covers a few tweaks with the output of the sentiment engine being saved off into a text file. 

from tweepy import Stream
from tweepy import OAuthHandler
from tweepy.streaming import StreamListener
import json
import sentiment_mod as s

#consumer key, consumer secret, access token, access secret.
ckey="*"
csecret="*"
atoken="*"
asecret="*"

class listener(StreamListener):

    def on_data(self, data):
        all_data = json.loads(data)
        tweet = all_data["text"]
        sentiment_value, confidence = s.sentiment(tweet)
        print(tweet, sentiment_value, confidence)

        if confidence*100 >= 80:
            output = open("twitter-out.txt", "a")
            output.write(sentiment_value)
            output.write('\n')
            output.close()

        return(True)

    def on_error(self, status):
        print(status)

auth = OAuthHandler(ckey, csecret)
auth.set_access_token(atoken, asecret)

twitterStream = Stream(auth, listener())
twitterStream.filter(track=["car"]) # term searched for in tweets

Next we’ll look at graphing this data.

Sentiment Analysis Module 

This section brings all the detail together to create a module that can be used to monitor twitter sentiment.

The code for this is seperated into two blocks.

The first pickles most of the heavy training to save time in future iteration:-.

import nltk
import random
#from nltk.corpus import movie_reviews
from nltk.classify.scikitlearn import SklearnClassifier
import pickle
from sklearn.naive_bayes import MultinomialNB, BernoulliNB
from sklearn.linear_model import LogisticRegression, SGDClassifier
from sklearn.svm import SVC, LinearSVC, NuSVC
from nltk.classify import ClassifierI
from statistics import mode
from nltk.tokenize import word_tokenize
#from unidecode import unidecode

class VoteClassifier(ClassifierI):
    def __init__(self, *classifiers):
        self._classifiers = classifiers

    def classify(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)
        return mode(votes)

    def confidence(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)

        choice_votes = votes.count(mode(votes))
        conf = choice_votes / len(votes)
        return conf

short_pos = open("positive.txt","r", encoding='utf-8', errors='replace').read() ## had to add a line to tell the open function to use utf-8
short_neg = open("negative.txt","r", encoding='utf-8', errors='replace').read()

# move this up here
all_words = []
documents = []

#  j is adject, r is adverb, and v is verb
#allowed_word_types = ["J","R","V"]
allowed_word_types = ["J"]

for p in short_pos.split('\n'):
    documents.append( (p, "pos") )
    words = word_tokenize(p)
    pos = nltk.pos_tag(words)
    for w in pos:
        if w[1][0] in allowed_word_types:
            all_words.append(w[0].lower())

for p in short_neg.split('\n'):
    documents.append( (p, "neg") )
    words = word_tokenize(p)
    pos = nltk.pos_tag(words)
    for w in pos:
        if w[1][0] in allowed_word_types:
            all_words.append(w[0].lower())

save_documents = open("pickled_algos/documents.pickle","wb")
pickle.dump(documents, save_documents)
save_documents.close()


all_words = nltk.FreqDist(all_words)


word_features = list(all_words.keys())[:5000]


save_word_features = open("pickled_algos/word_features5k.pickle","wb")
pickle.dump(word_features, save_word_features)
save_word_features.close()


def find_features(document):
    words = word_tokenize(document)
    features = {}
    for w in word_features:
        features[w] = (w in words)

    return features

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

random.shuffle(featuresets)
print(len(featuresets))

testing_set = featuresets[7400:] # memory limitations on the Pi meant this needed reducing from 10,000
training_set = featuresets[:7400]

save_featuresets = open("pickled_algos/featuresets.pickle","wb") ## added code to pickle featuresets
pickle.dump(featuresets, save_featuresets)
save_featuresets.close()

classifier = nltk.NaiveBayesClassifier.train(training_set)
print("Original Naive Bayes Algo accuracy percent:", (nltk.classify.accuracy(classifier, testing_set))*100)
classifier.show_most_informative_features(15)

###############
save_classifier = open("pickled_algos/originalnaivebayes5k.pickle","wb")
pickle.dump(classifier, save_classifier)
save_classifier.close()

MNB_classifier = SklearnClassifier(MultinomialNB())
MNB_classifier.train(training_set)
print("MNB_classifier accuracy percent:", (nltk.classify.accuracy(MNB_classifier, testing_set))*100)

save_classifier = open("pickled_algos/MNB_classifier5k.pickle","wb")
pickle.dump(MNB_classifier, save_classifier)
save_classifier.close()

BernoulliNB_classifier = SklearnClassifier(BernoulliNB())
BernoulliNB_classifier.train(training_set)
print("BernoulliNB_classifier accuracy percent:", (nltk.classify.accuracy(BernoulliNB_classifier, testing_set))*100)

save_classifier = open("pickled_algos/BernoulliNB_classifier5k.pickle","wb")
pickle.dump(BernoulliNB_classifier, save_classifier)
save_classifier.close()

LogisticRegression_classifier = SklearnClassifier(LogisticRegression())
LogisticRegression_classifier.train(training_set)
print("LogisticRegression_classifier accuracy percent:", (nltk.classify.accuracy(LogisticRegression_classifier, testing_set))*100)

save_classifier = open("pickled_algos/LogisticRegression_classifier5k.pickle","wb")
pickle.dump(LogisticRegression_classifier, save_classifier)
save_classifier.close()


LinearSVC_classifier = SklearnClassifier(LinearSVC())
LinearSVC_classifier.train(training_set)
print("LinearSVC_classifier accuracy percent:", (nltk.classify.accuracy(LinearSVC_classifier, testing_set))*100)

save_classifier = open("pickled_algos/LinearSVC_classifier5k.pickle","wb")
pickle.dump(LinearSVC_classifier, save_classifier)
save_classifier.close()


##NuSVC_classifier = SklearnClassifier(NuSVC())
##NuSVC_classifier.train(training_set)
##print("NuSVC_classifier accuracy percent:", (nltk.classify.accuracy(NuSVC_classifier, testing_set))*100)


SGDC_classifier = SklearnClassifier(SGDClassifier())
SGDC_classifier.train(training_set)
print("SGDClassifier accuracy percent:",nltk.classify.accuracy(SGDC_classifier, testing_set)*100)

save_classifier = open("pickled_algos/SGDC_classifier5k.pickle","wb")
pickle.dump(SGDC_classifier, save_classifier)
save_classifier.close()

I had to drop the size of the learning set given the memory restrictions on the Pi not sure there was another way around this and it probably means that to start doing some of this stuff seriously….I’m going to need a bigger/better server.

Also I added a section to pickle the featuresets:-

save_featuresets = open("pickled_algos/featuresets.pickle","wb") ## added code to pickle featuresets
pickle.dump(featuresets, save_featuresets)
save_featuresets.close()

One option might be clustering the Pi or putting together a low power server – which then leads into building a full home server farm.

The next section is then the sentiment module:-

#File: sentiment_mod.py

import nltk
import random
#from nltk.corpus import movie_reviews
from nltk.classify.scikitlearn import SklearnClassifier
import pickle
from sklearn.naive_bayes import MultinomialNB, BernoulliNB
from sklearn.linear_model import LogisticRegression, SGDClassifier
from sklearn.svm import SVC, LinearSVC, NuSVC
from nltk.classify import ClassifierI
from statistics import mode
from nltk.tokenize import word_tokenize



class VoteClassifier(ClassifierI):
    def __init__(self, *classifiers):
        self._classifiers = classifiers

    def classify(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)
        return mode(votes)

    def confidence(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)

        choice_votes = votes.count(mode(votes))
        conf = choice_votes / len(votes)
        return conf


documents_f = open("pickled_algos/documents.pickle", "rb")
documents = pickle.load(documents_f)
documents_f.close()




word_features5k_f = open("pickled_algos/word_features5k.pickle", "rb")
word_features = pickle.load(word_features5k_f)
word_features5k_f.close()


def find_features(document):
    words = word_tokenize(document)
    features = {}
    for w in word_features:
        features[w] = (w in words)

    return features



featuresets_f = open("pickled_algos/featuresets.pickle", "rb")
featuresets = pickle.load(featuresets_f)
featuresets_f.close()

random.shuffle(featuresets)
print(len(featuresets))

testing_set = featuresets[10000:]
training_set = featuresets[:10000]



open_file = open("pickled_algos/originalnaivebayes5k.pickle", "rb")
classifier = pickle.load(open_file)
open_file.close()


open_file = open("pickled_algos/MNB_classifier5k.pickle", "rb")
MNB_classifier = pickle.load(open_file)
open_file.close()



open_file = open("pickled_algos/BernoulliNB_classifier5k.pickle", "rb")
BernoulliNB_classifier = pickle.load(open_file)
open_file.close()


open_file = open("pickled_algos/LogisticRegression_classifier5k.pickle", "rb")
LogisticRegression_classifier = pickle.load(open_file)
open_file.close()


open_file = open("pickled_algos/LinearSVC_classifier5k.pickle", "rb")
LinearSVC_classifier = pickle.load(open_file)
open_file.close()


open_file = open("pickled_algos/SGDC_classifier5k.pickle", "rb")
SGDC_classifier = pickle.load(open_file)
open_file.close()

voted_classifier = VoteClassifier(
                                  classifier,
                                  LinearSVC_classifier,
                                  MNB_classifier,
                                  BernoulliNB_classifier,
                                  LogisticRegression_classifier)


def sentiment(text):
    feats = find_features(text)
    return voted_classifier.classify(feats),voted_classifier.confidence(feats)

Better training data

This tutorial covers training the algorithm on a new more tailored data-set. The training data set used still covers movie reviews but contains those that are a lot shorter – which should give better results.

Investigating Bias

Combining Algorithms with NLTK

In this section we combine the classifiers from previous steps into a voting classifier that will give an aggregated confidence level. Each algorithm gets one vote, and the classification that has the votes votes is the chosen one.

To do this we’ll import the modules we need:-

from nltk.classify import ClassifierI
from statistics import mode # to allow us to add up and compare votes

Next build a VoteClassifier class:-

class VoteClassifier(ClassifierI):
    def __init__(self, *classifiers):
        self._classifiers = classifiers
    
    def classify(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)
        return mode(votes)
    
    def confidence(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)

        choice_votes = votes.count(mode(votes))
        conf = choice_votes / len(votes)
        return conf

And then combine it all:-

import nltk
import random
from nltk.corpus import movie_reviews
from nltk.classify.scikitlearn import SklearnClassifier
import pickle

from sklearn.naive_bayes import MultinomialNB, BernoulliNB
from sklearn.linear_model import LogisticRegression, SGDClassifier
from sklearn.svm import SVC, LinearSVC, NuSVC

from nltk.classify import ClassifierI
from statistics import mode


class VoteClassifier(ClassifierI):
    def __init__(self, *classifiers):
        self._classifiers = classifiers

    def classify(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)
        return mode(votes)

    def confidence(self, features):
        votes = []
        for c in self._classifiers:
            v = c.classify(features)
            votes.append(v)

        choice_votes = votes.count(mode(votes))
        conf = choice_votes / len(votes)
        return conf

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)

word_features = list(all_words.keys())[:3000]

def find_features(document):
    words = set(document)
    features = {}
    for w in word_features:
        features[w] = (w in words)

    return features

#print((find_features(movie_reviews.words('neg/cv000_29416.txt'))))

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

#classifier = nltk.NaiveBayesClassifier.train(training_set)

classifier_f = open("naivebayes.pickle","rb")
classifier = pickle.load(classifier_f)
classifier_f.close()




print("Original Naive Bayes Algo accuracy percent:", (nltk.classify.accuracy(classifier, testing_set))*100)
classifier.show_most_informative_features(15)

MNB_classifier = SklearnClassifier(MultinomialNB())
MNB_classifier.train(training_set)
print("MNB_classifier accuracy percent:", (nltk.classify.accuracy(MNB_classifier, testing_set))*100)

BernoulliNB_classifier = SklearnClassifier(BernoulliNB())
BernoulliNB_classifier.train(training_set)
print("BernoulliNB_classifier accuracy percent:", (nltk.classify.accuracy(BernoulliNB_classifier, testing_set))*100)

LogisticRegression_classifier = SklearnClassifier(LogisticRegression())
LogisticRegression_classifier.train(training_set)
print("LogisticRegression_classifier accuracy percent:", (nltk.classify.accuracy(LogisticRegression_classifier, testing_set))*100)

SGDClassifier_classifier = SklearnClassifier(SGDClassifier())
SGDClassifier_classifier.train(training_set)
print("SGDClassifier_classifier accuracy percent:", (nltk.classify.accuracy(SGDClassifier_classifier, testing_set))*100)

##SVC_classifier = SklearnClassifier(SVC())
##SVC_classifier.train(training_set)
##print("SVC_classifier accuracy percent:", (nltk.classify.accuracy(SVC_classifier, testing_set))*100)

LinearSVC_classifier = SklearnClassifier(LinearSVC())
LinearSVC_classifier.train(training_set)
print("LinearSVC_classifier accuracy percent:", (nltk.classify.accuracy(LinearSVC_classifier, testing_set))*100)

NuSVC_classifier = SklearnClassifier(NuSVC())
NuSVC_classifier.train(training_set)
print("NuSVC_classifier accuracy percent:", (nltk.classify.accuracy(NuSVC_classifier, testing_set))*100)


voted_classifier = VoteClassifier(classifier,
                                  NuSVC_classifier,
                                  LinearSVC_classifier,
                                  SGDClassifier_classifier,
                                  MNB_classifier,
                                  BernoulliNB_classifier,
                                  LogisticRegression_classifier)

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

print("Classification:", voted_classifier.classify(testing_set[0][0]), "Confidence %:",voted_classifier.confidence(testing_set[0][0])*100)
print("Classification:", voted_classifier.classify(testing_set[1][0]), "Confidence %:",voted_classifier.confidence(testing_set[1][0])*100)
print("Classification:", voted_classifier.classify(testing_set[2][0]), "Confidence %:",voted_classifier.confidence(testing_set[2][0])*100)
print("Classification:", voted_classifier.classify(testing_set[3][0]), "Confidence %:",voted_classifier.confidence(testing_set[3][0])*100)
print("Classification:", voted_classifier.classify(testing_set[4][0]), "Confidence %:",voted_classifier.confidence(testing_set[4][0])*100)
print("Classification:", voted_classifier.classify(testing_set[5][0]), "Confidence %:",voted_classifier.confidence(testing_set[5][0])*100)

Output:-

Original Naive Bayes Algo accuracy percent: 66.0 
Most Informative Features 
    thematic = True pos : neg = 9.1 : 1.0 
    secondly = True pos : neg = 8.5 : 1.0 
    narrates = True pos : neg = 7.8 : 1.0 
    layered = True pos : neg = 7.1 : 1.0 
    rounded = True pos : neg = 7.1 : 1.0 
    supreme = True pos : neg = 7.1 : 1.0 
    crappy = True neg : pos = 6.9 : 1.0 
    uplifting = True pos : neg = 6.2 : 1.0 
    ugh = True neg : pos = 5.3 : 1.0 
    gaining = True pos : neg = 5.1 : 1.0 
    mamet = True pos : neg = 5.1 : 1.0 
    wanda = True neg : pos = 4.9 : 1.0 
    onset = True neg : pos = 4.9 : 1.0 
    fantastic = True pos : neg = 4.5 : 1.0 
    milos = True pos : neg = 4.4 : 1.0 
MNB_classifier accuracy percent: 67.0 
BernoulliNB_classifier accuracy percent: 67.0 LogisticRegression_classifier accuracy percent: 68.0 SGDClassifier_classifier accuracy percent: 57.99999999999999 LinearSVC_classifier accuracy percent: 67.0 
NuSVC_classifier accuracy percent: 65.0 
voted_classifier accuracy percent: 65.0 
Classification: neg Confidence %: 100.0 
Classification: pos Confidence %: 57.14285714285714 
Classification: neg Confidence %: 57.14285714285714 
Classification: neg Confidence %: 57.14285714285714 
Classification: pos Confidence %: 57.14285714285714 
Classification: pos Confidence %: 85.71428571428571

One issue that did crop up on my system was the error:-

/usr/local/lib/python3.7/dist-packages/sklearn/linear_model/_logistic.py:940: ConvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression
extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)

Scikit-Learn Sklearn

The Scikit-learn (sklearn) module adds a bunch of other classifiers as well as machine learning algorithms – so although NLTK has Naive Bayes algorithms it’s straight forward to incorporate others from SciKit Learn.

As with most Python modules this involves importing the various parts before using them.

To import:-

from nltk.classify.scikitlearn import SklearnClassifier
from sklearn.naive_bayes import MultinomialNB, GaussianNB, BernoulliNB
from sklearn.linear_model import LogisticRegression, SGDClassifier
from sklearn.svm    import SVC, LinearSVC, NuSVC

Then using the classifiers:-

## Multinomial Naive Bayse
MNB_classifier = SklearnClassifier(MultinomialNB())
MNB_classifier.train(training_set)
print("MNB_classifier accuracy:", (nltk.classify.accuracy(MNB_classifier, testing_set))*100)

## Gaussian Naive Bayse
#GNB_classifier = SklearnClassifier(GaussianNB())
#GNB_classifier.train(training_set)
#print("GNB_classifier accuracy:", (nltk.classify.accuracy(GNB_classifier, testing_set))*100)

## Bernoulli Naive Bayse
BNB_classifier = SklearnClassifier(BernoulliNB())
BNB_classifier.train(training_set)
print("BNB_classifier accuracy:", (nltk.classify.accuracy(BNB_classifier, testing_set))*100)

LogisticRegression_classifier = SklearnClassifier(LogisticRegression())
LogisticRegression_classifier.train(training_set)
print("LogisticRegression_classifier accuracy percent:", (nltk.classify.accuracy(LogisticRegression_classifier, testing_set))*100)

SGDClassifier_classifier = SklearnClassifier(SGDClassifier())
SGDClassifier_classifier.train(training_set)
print("SGDClassifier_classifier accuracy percent:", (nltk.classify.accuracy(SGDClassifier_classifier, testing_set))*100)

SVC_classifier = SklearnClassifier(SVC())
SVC_classifier.train(training_set)
print("SVC_classifier accuracy percent:", (nltk.classify.accuracy(SVC_classifier, testing_set))*100)

LinearSVC_classifier = SklearnClassifier(LinearSVC())
LinearSVC_classifier.train(training_set)
print("LinearSVC_classifier accuracy percent:", (nltk.classify.accuracy(LinearSVC_classifier, testing_set))*100)

NuSVC_classifier = SklearnClassifier(NuSVC())
NuSVC_classifier.train(training_set)
print("NuSVC_classifier accuracy percent:", (nltk.classify.accuracy(NuSVC_classifier, testing_set))*100)

This gives a range of different views on +ve/-ve and the next section will focus on a voting system for these.

Pickle – saving trained classifiers

Once a classifier has been trained it’s often quicker in the long run to save the algorithm for re-use later, rather than training it each and every time.

Script to Save the classifier:-

save_classifier = open("naivebayes.pickle","wb")
pickle.dump(classifier, save_classifier)
save_classifier.close()

Script to load the classifier once pickled:-

classifier_f = open("naivebayes.pickle", "rb")
classifier = pickle.load(classifier_f)
classifier_f.close()

import nltk
import random
import pickle
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 = []
for w in movie_reviews.words():
    all_words.append(w.lower()) # normalise everything to lower case and append

all_words = nltk.FreqDist(all_words) # converts to a nltk frequency distribution

word_features = list (all_words.keys())[:3000] # from the frequency list we're taking just the words(keys) and only the top 3000

def find_fetures(document):
    words = set(document) # this gives a list of the unique words - removes duplicates
    features = {} # declare an empty dictionary
    for w in word_features:
        features[w] = (w in words) # this checks each word in the top 3000 to see if it is present in the passed text 'document' so gives a true/false against the 3000
    return features

# print((find_fetures(movie_reviews.words('neg/cv000_29416.txt'))))

featuresets = [(find_fetures(rev), category) for (rev, category) in documents]

training_set = featuresets[:1900] # splits the featuresets into two seperate groups 1 to train and the other to test
testing_set = featuresets[1900:]

## Naive Bayse Algorythm

# classifier = nltk.NaiveBayesClassifier.train(training_set) # training the NaiveBayesClassifier on training data commented out once naivebayes.pickle is generated
classifier_f = open("naivebayes.pickle","rb")
classifer = pickle.load(classifier_f)
classifier_f.close()

print("Naive Bayes Algo accuracy:", (nltk.classify.accuracy(classifier, testing_set))*100)
classifier.show_most_informative_features(15) # tells most popular words on both sides and if +ve or -ve

## Pickle allows you to save python obects and import them later
# save_classifier = open("naivebayes.pickle","wb") # commented out once naivebayes.pickle has been generated
# pickle.dump(classifier, save_classifier)
# save_classifier.close()

Naive Bayes

This section builds on the last 2 tutorials to choose an algorithm, separate the data into training and testing sets – and set it running.

The algorithm in this example is the Naive Bayes classifier.

But first the data needs to be split into training and test sets for some supervised machine learning. In essence we show the machine data, and telling it “this data is positive,” or “this data is negative.” Then, after the training is done, we show the machine some new data and ask the computer what the computer thinks the category of the new data is.

import nltk
import random
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 = []
for w in movie_reviews.words():
    all_words.append(w.lower()) # normalise everything to lower case and append

all_words = nltk.FreqDist(all_words) # converts to a nltk frequency distribution

word_features = list (all_words.keys())[:3000] # from the frequency list we're taking just the words(keys) and only the top 3000

def find_fetures(document):
    words = set(document) # this gives a list of the unique words - removes duplicates
    features = {} # declare an empty dictionary
    for w in word_features:
        features[w] = (w in words) # this checks each word in the top 3000 to see if it is present in the passed text 'document' so gives a true/false against the 3000
    return features

# print((find_fetures(movie_reviews.words('neg/cv000_29416.txt'))))

featuresets = [(find_fetures(rev), category) for (rev, category) in documents]

training_set = featuresets[:1900] # splits the featuresets into two seperate groups 1 to train and the other to test
testing_set = featuresets[1900:]

## Naive Bayse Algorythm

classifier = nltk.NaiveBayesClassifier.train(training_set) # training the NaiveBayesClassifier on training data
print("Naive Bayes Algo accuracy:", (nltk.classify.accuracy(classifier, testing_set))*100)
classifier.show_most_informative_features(15) # tells most popular words on both sides and if +ve or -ve

Output:-

galiquis@raspberrypi: $ python3 ./nltk_tutorial13.py
Naive Bayes Algo accuracy: 80.0
Most Informative Features
annual = True pos : neg = 9.6 : 1.0
sucks = True neg : pos = 9.1 : 1.0
bothered = True neg : pos = 9.1 : 1.0
frances = True pos : neg = 8.9 : 1.0
idiotic = True neg : pos = 8.8 : 1.0
unimaginative = True neg : pos = 8.4 : 1.0
silverstone = True neg : pos = 7.7 : 1.0
shoddy = True neg : pos = 7.1 : 1.0
suvari = True neg : pos = 7.1 : 1.0
mena = True neg : pos = 7.1 : 1.0
sexist = True neg : pos = 7.1 : 1.0
regard = True pos : neg = 6.9 : 1.0
schumacher = True neg : pos = 6.7 : 1.0
uninspired = True neg : pos = 6.6 : 1.0
kidding = True neg : pos = 6.4 : 1.0

Words as Features for Learning

In order to apply some machine learning to the algorithm we need to identify features in the date (word).

This section covers compiling feature lists from positive and negative  reviews, to hopefully see trends.

import nltk
import random
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 = []
for w in movie_reviews.words():
    all_words.append(w.lower()) # normalise everything to lower case and append

all_words = nltk.FreqDist(all_words) # converts to a nltk frequency distribution

word_features = list (all_words.keys())[:3000] # from the frequency list we're taking just the words(keys) and only the top 3000

def find_fetures(document):
    words = set(document) # this gives a list of the unique words - removes duplicates
    features = {} # declare an empty dictionary
    for w in word_features:
        features[w] = (w in words) # this checks each word in the top 3000 to see if it is present in the passed text 'document' so gives a true/false against the 3000
    return features

print((find_fetures(movie_reviews.words('neg/cv000_29416.txt'))))

featuresets = [(find_fetures(rev), category) for (rev, category) in documents]

Text Classification

The step on from looking at words and how words relate to each other, is to broaden out to classifying sections of text. Classification can be focused on identifying what a piece of text is about (e.g. politics, the military etc) or as simple as identifying if some text is spam or not spam – for comment/email filters.

import nltk
import random
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)

print(documents[1])

all_words = []
for w in movie_reviews.words():
    all_words.append(w.lower()) # normalise everything to lower case and append

all_words = nltk.FreqDist(all_words) # converts to a nltk frequency distribution
print(all_words.most_common(15)) # top 15 most common
print(all_words["stupid"]) # shows the frequency of a specific word

The first key piece of code is defining ‘documents’.

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

This uses the list() constructor to iterate through the movie_reviews and create a tuple. 

However, this is easier to see when broken down into multiple lines:-

documents = [] # declare the list
for category in movie_reviews.categories():
    for fileid in movie_reviews.fileids(category):
        documents.append(list(movie_reviews.words(fileid)), category)

Next we shuffle the documents as we don’t want to train and test on the same data.

So the principal being used is to take all of the words from all of the reviews and compile them to find out the most popular words and whether they appear in positive or negative reviews. Then when we look at a new review we can test what words appear and determine if they are more positive or negative….onto the next tutorial.

Useful links:-

Lists and Tuples – https://realpython.com/python-lists-tuples/

Wordnet

WordNet is a lexical database for the English language, created by Princeton, and is part of the NLTK corpus.

WordNet can be used alongside NLTK to find the meanings of words, synonyms, antonyms, context, etc.

Below is an example of calling various aspects of Synsets for the word program:-

from nltk.corpus import wordnet

syns = wordnet.synsets("program")
print(syns)

print(syns[0])
#synset 0
print(syns[0].name())

# just the word
print(syns[0].lemmas()[0].name())

# definition
print(syns[0].definition())

# examples
print(syns[0].examples())

So Synsets has a number of entries for ‘program’:-

• [Synset(‘plan.n.01’
• Synset(‘program.n.02’)
• Synset(‘broadcast.n.02’)
• Synset(‘platform.n.02’)
• Synset(‘program.n.05’)
• Synset(‘course_of_study.n.01’)
• Synset(‘program.n.07’)
• Synset(‘program.n.08’),
• Synset(‘program.v.01’),
• Synset(‘program.v.02’)]

All relating to different meanings for ‘program’.

Once a Synset is selected its Lemma, definition or examples can be accessed.

Synonyms & Antonyms

Synsets can also be used to generate Synonyms and Antonyms for a given word, in the below example “Good”…

from nltk.corpus import wordnet

synonyms = [] #declare an empty list
antonyms = [] #declare an empty list

for syn in wordnet.synsets("good"):
    for lemma in syn.lemmas():
        synonyms.append(lemma.name())
        if lemma.antonyms():
            antonyms.append(lemma.antonyms()[0].name())

print("Synonyms",set(synonyms))
print("Antonyms",set(antonyms))

Giving the output:-

Synonyms {‘respectable’, ‘salutary’, ‘effective’, ‘dependable’, ‘thoroughly’, ‘beneficial’, ‘unspoilt’, ‘secure’, ‘good’, ‘skilful’, ‘in_effect’, ‘well’, ‘dear’, ‘undecomposed’, ‘serious’, ‘soundly’, ‘goodness’, ‘practiced’, ‘honorable’, ‘safe’, ‘expert’, ‘in_force’, ‘honest’, ‘sound’, ‘full’, ‘trade_good’, ‘skillful’, ‘ripe’, ‘upright’, ‘just’, ‘right’, ‘estimable’, ‘near’, ‘adept’, ‘unspoiled’, ‘proficient’, ‘commodity’}

Antonyms {‘evilness’, ‘evil’, ‘badness’, ‘ill’, ‘bad’}

Semantic similarities

WordNet can also be used to compare the similarity of two words and their tenses, by using the Wu and Palmer method for semantic comparison.

from nltk.corpus import wordnet

w1 = wordnet.synset("ship.n.01") # word.(n)oun.(1)st_occurance
w2 = wordnet.synset("boat.n.01")
print(w1.wup_similarity(w2)) # apllying the Wu & Palmer method

w1 = wordnet.synset("ship.n.01") # word.(n)oun.(1)st_occurance
w2 = wordnet.synset("car.n.01")
print(w1.wup_similarity(w2)) # apllying the Wu & Palmer method

w1 = wordnet.synset("ship.n.01") # word.(n)oun.(1)st_occurance
w2 = wordnet.synset("cat.n.01")
print(w1.wup_similarity(w2)) # apllying the Wu & Palmer method

NLTK Corpa

This tutorial just covers navigating the Corpa.

I used the following to find the data directory:-

galiquis@raspberrypi:~ $ sudo find / -type d -name “nltk_data”

Lemmatizing

Lemmatizing is very similar to stemming with the key difference being that lemmatizing ends up at a real word.

from nltk.stem import WordNetLemmatizer

lemmatizer = WordNetLemmatizer()
print(lemmatizer.lemmatize("cats"))
print(lemmatizer.lemmatize("cacti"))
print(lemmatizer.lemmatize("geese"))
print(lemmatizer.lemmatize("rocks"))
print(lemmatizer.lemmatize("python"))
print(lemmatizer.lemmatize("better", pos="a"))
print(lemmatizer.lemmatize("best", pos="a"))
print(lemmatizer.lemmatize("run"))
print(lemmatizer.lemmatize("run",'v'))

Gives the output:-

galiquis@raspberrypi:$ python3 ./nltk_tutorial8.py
cat
cactus
goose
rock
python
good
best
run
run

Some points to note:-

• Lemmatize takes part of the POS parameter/tag so:-
  • pos=”a”or ‘a’ will find the closest adjective
  • pos=”v” or ‘v’ will find the closest verb
  • the default (no option) finds the closest noun
• More powerful than stemming

Named Entity Recognition

A key method for chunking in natural language processing is called “Named Entity Recognition.” The concept being to have the code identify and pull out “entities” like people, places, things, locations, monetary figures, and more.

NE Type and Examples
ORGANIZATION – Georgia-Pacific Corp., WHO
PERSON – Eddy Bonte, President Obama
LOCATION – Murray River, Mount Everest
DATE – June, 2008-06-29
TIME – two fifty a m, 1:30 p.m.
MONEY – 175 million Canadian Dollars, GBP 10.40
PERCENT – twenty pct, 18.75 %
FACILITY – Washington Monument, Stonehenge
GPE – South East Asia, Midlothian

import nltk
from nltk.corpus import state_union
from nltk.tokenize import PunktSentenceTokenizer

train_text = state_union.raw("2005-GWBush.txt")
sample_text = state_union.raw("2006-GWBush.txt")

custom_sent_tokenizer = PunktSentenceTokenizer(train_text)

tokenized = custom_sent_tokenizer.tokenize(sample_text)

def process_content():
    try:
        for i in tokenized[5:]:
            words = nltk.word_tokenize(i)
            tagged = nltk.pos_tag(words)
            namedEnt = nltk.ne_chunk(tagged, binary=True)
            namedEnt.draw()
    except Exception as e:
        print(str(e))

process_content()

The key line that applies the entity recognition is:-

namedEnt = nltk.ne_chunk(tagged, binary=True)

Which has two options…

• binary=True – this means either something is a named entity, or not. There will be no further detail and hits are titled ‘NE’
• binary=False – this provides detail on the named entity, e.g Facility, Organisation, Person etc 

In the example when Binary is False, it picked up the same tagged items, but split up terms like White House into “White” and “House” as if they were different, whereas in the binary = True option, the named entity recognition was correct to say White House was part of the same named entity.

NLTK’s Named Entity Recognition can be a bit hit and miss leading to a lot of false positives.