-
Notifications
You must be signed in to change notification settings - Fork 47
/
positivenaivebayes.html
319 lines (233 loc) · 22.8 KB
/
positivenaivebayes.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
<!DOCTYPE html>
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta name="theme-color" content="#2D2D2D" />
<title>NLTK :: nltk.classify.positivenaivebayes</title>
<link rel="stylesheet" href="../../../_static/css/nltk_theme.css"/>
<link rel="stylesheet" href="../../../_static/css/custom.css"/>
<script type="text/javascript" id="documentation_options" data-url_root="../../../" src="../../../_static/documentation_options.js"></script>
<script type="text/javascript" src="../../../_static/documentation_options.js"></script>
<script type="text/javascript" src="../../../_static/jquery.js"></script>
<script type="text/javascript" src="../../../_static/underscore.js"></script>
<script type="text/javascript" src="../../../_static/doctools.js"></script>
<script src="https://email.tl.fortawesome.com/c/eJxNjUEOgyAQAF8jR7Kw6wIHDh7sP1Cw2mgxgmn6-3JsMqc5zEQfE8dkxOY1KKMUOI3ACFKRJpSW2AAp7ontYIaxI6i7XPJVwyeVfCQ550Os3jLrGSNOLgbdAy6s0PBk2TFNjEbsfq31LB0OnX407pJa5v2faRadwSW63mn5KuLyR9j2tgx3zecanl-55R_-jjPs"></script>
</head>
<body>
<div id="nltk-theme-container">
<header>
<div id="logo-container">
<h1>
<a href="../../../index.html">NLTK</a>
</h1>
</div>
<div id="project-container">
<h1>Documentation</h1>
</div>
<a id="menu-toggle" class="fa fa-bars" aria-hidden="true"></a>
<script type="text/javascript">
$("#menu-toggle").click(function() {
$("#menu-toggle").toggleClass("toggled");
$("#side-menu-container").slideToggle(300);
});
</script>
</header>
<div id="content-container">
<div id="side-menu-container">
<div id="search" role="search">
<form id="rtd-search-form" class="wy-form" action="../../../search.html" method="get">
<input type="text" name="q" placeholder="Search" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
<div id="side-menu" role="navigation">
<p class="caption" role="heading"><span class="caption-text">NLTK Documentation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../api/nltk.html">API Reference</a></li>
<li class="toctree-l1"><a class="reference external" href="http://www.nltk.org/howto">Example Usage</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../py-modindex.html">Module Index</a></li>
<li class="toctree-l1"><a class="reference external" href="https://github.com/nltk/nltk/wiki">Wiki</a></li>
<li class="toctree-l1"><a class="reference external" href="https://github.com/nltk/nltk/wiki/FAQ">FAQ</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Installation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../install.html">Installing NLTK</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../data.html">Installing NLTK Data</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">More</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../news.html">Release Notes</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../contribute.html">Contributing to NLTK</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../team.html">NLTK Team</a></li>
</ul>
</div>
</div>
<div id="main-content-container">
<div id="main-content" role="main">
<h1>Source code for nltk.classify.positivenaivebayes</h1><div class="highlight"><pre>
<span></span><span class="c1"># Natural Language Toolkit: Positive Naive Bayes Classifier</span>
<span class="c1">#</span>
<span class="c1"># Copyright (C) 2012 NLTK Project</span>
<span class="c1"># Author: Alessandro Presta <alessandro.presta@gmail.com></span>
<span class="c1"># URL: <http://nltk.org/></span>
<span class="c1"># For license information, see LICENSE.TXT</span>
<span class="sd">"""</span>
<span class="sd">A variant of the Naive Bayes Classifier that performs binary classification with</span>
<span class="sd">partially-labeled training sets. In other words, assume we want to build a classifier</span>
<span class="sd">that assigns each example to one of two complementary classes (e.g., male names and</span>
<span class="sd">female names).</span>
<span class="sd">If we have a training set with labeled examples for both classes, we can use a</span>
<span class="sd">standard Naive Bayes Classifier. However, consider the case when we only have labeled</span>
<span class="sd">examples for one of the classes, and other, unlabeled, examples.</span>
<span class="sd">Then, assuming a prior distribution on the two labels, we can use the unlabeled set</span>
<span class="sd">to estimate the frequencies of the various features.</span>
<span class="sd">Let the two possible labels be 1 and 0, and let's say we only have examples labeled 1</span>
<span class="sd">and unlabeled examples. We are also given an estimate of P(1).</span>
<span class="sd">We compute P(feature|1) exactly as in the standard case.</span>
<span class="sd">To compute P(feature|0), we first estimate P(feature) from the unlabeled set (we are</span>
<span class="sd">assuming that the unlabeled examples are drawn according to the given prior distribution)</span>
<span class="sd">and then express the conditional probability as:</span>
<span class="sd">| P(feature) - P(feature|1) * P(1)</span>
<span class="sd">| P(feature|0) = ----------------------------------</span>
<span class="sd">| P(0)</span>
<span class="sd">Example:</span>
<span class="sd"> >>> from nltk.classify import PositiveNaiveBayesClassifier</span>
<span class="sd">Some sentences about sports:</span>
<span class="sd"> >>> sports_sentences = [ 'The team dominated the game',</span>
<span class="sd"> ... 'They lost the ball',</span>
<span class="sd"> ... 'The game was intense',</span>
<span class="sd"> ... 'The goalkeeper catched the ball',</span>
<span class="sd"> ... 'The other team controlled the ball' ]</span>
<span class="sd">Mixed topics, including sports:</span>
<span class="sd"> >>> various_sentences = [ 'The President did not comment',</span>
<span class="sd"> ... 'I lost the keys',</span>
<span class="sd"> ... 'The team won the game',</span>
<span class="sd"> ... 'Sara has two kids',</span>
<span class="sd"> ... 'The ball went off the court',</span>
<span class="sd"> ... 'They had the ball for the whole game',</span>
<span class="sd"> ... 'The show is over' ]</span>
<span class="sd">The features of a sentence are simply the words it contains:</span>
<span class="sd"> >>> def features(sentence):</span>
<span class="sd"> ... words = sentence.lower().split()</span>
<span class="sd"> ... return dict(('contains(%s)' % w, True) for w in words)</span>
<span class="sd">We use the sports sentences as positive examples, the mixed ones ad unlabeled examples:</span>
<span class="sd"> >>> positive_featuresets = map(features, sports_sentences)</span>
<span class="sd"> >>> unlabeled_featuresets = map(features, various_sentences)</span>
<span class="sd"> >>> classifier = PositiveNaiveBayesClassifier.train(positive_featuresets,</span>
<span class="sd"> ... unlabeled_featuresets)</span>
<span class="sd">Is the following sentence about sports?</span>
<span class="sd"> >>> classifier.classify(features('The cat is on the table'))</span>
<span class="sd"> False</span>
<span class="sd">What about this one?</span>
<span class="sd"> >>> classifier.classify(features('My team lost the game'))</span>
<span class="sd"> True</span>
<span class="sd">"""</span>
<span class="kn">from</span> <span class="nn">collections</span> <span class="kn">import</span> <span class="n">defaultdict</span>
<span class="kn">from</span> <span class="nn">nltk.classify.naivebayes</span> <span class="kn">import</span> <span class="n">NaiveBayesClassifier</span>
<span class="kn">from</span> <span class="nn">nltk.probability</span> <span class="kn">import</span> <span class="n">DictionaryProbDist</span><span class="p">,</span> <span class="n">ELEProbDist</span><span class="p">,</span> <span class="n">FreqDist</span>
<span class="c1">##//////////////////////////////////////////////////////</span>
<span class="c1">## Positive Naive Bayes Classifier</span>
<span class="c1">##//////////////////////////////////////////////////////</span>
<div class="viewcode-block" id="PositiveNaiveBayesClassifier"><a class="viewcode-back" href="../../../api/nltk.classify.positivenaivebayes.html#nltk.classify.positivenaivebayes.PositiveNaiveBayesClassifier">[docs]</a><span class="k">class</span> <span class="nc">PositiveNaiveBayesClassifier</span><span class="p">(</span><span class="n">NaiveBayesClassifier</span><span class="p">):</span>
<div class="viewcode-block" id="PositiveNaiveBayesClassifier.train"><a class="viewcode-back" href="../../../api/nltk.classify.positivenaivebayes.html#nltk.classify.positivenaivebayes.PositiveNaiveBayesClassifier.train">[docs]</a> <span class="nd">@staticmethod</span>
<span class="k">def</span> <span class="nf">train</span><span class="p">(</span>
<span class="n">positive_featuresets</span><span class="p">,</span>
<span class="n">unlabeled_featuresets</span><span class="p">,</span>
<span class="n">positive_prob_prior</span><span class="o">=</span><span class="mf">0.5</span><span class="p">,</span>
<span class="n">estimator</span><span class="o">=</span><span class="n">ELEProbDist</span><span class="p">,</span>
<span class="p">):</span>
<span class="sd">"""</span>
<span class="sd"> :param positive_featuresets: An iterable of featuresets that are known as positive</span>
<span class="sd"> examples (i.e., their label is ``True``).</span>
<span class="sd"> :param unlabeled_featuresets: An iterable of featuresets whose label is unknown.</span>
<span class="sd"> :param positive_prob_prior: A prior estimate of the probability of the label</span>
<span class="sd"> ``True`` (default 0.5).</span>
<span class="sd"> """</span>
<span class="n">positive_feature_freqdist</span> <span class="o">=</span> <span class="n">defaultdict</span><span class="p">(</span><span class="n">FreqDist</span><span class="p">)</span>
<span class="n">unlabeled_feature_freqdist</span> <span class="o">=</span> <span class="n">defaultdict</span><span class="p">(</span><span class="n">FreqDist</span><span class="p">)</span>
<span class="n">feature_values</span> <span class="o">=</span> <span class="n">defaultdict</span><span class="p">(</span><span class="nb">set</span><span class="p">)</span>
<span class="n">fnames</span> <span class="o">=</span> <span class="nb">set</span><span class="p">()</span>
<span class="c1"># Count up how many times each feature value occurred in positive examples.</span>
<span class="n">num_positive_examples</span> <span class="o">=</span> <span class="mi">0</span>
<span class="k">for</span> <span class="n">featureset</span> <span class="ow">in</span> <span class="n">positive_featuresets</span><span class="p">:</span>
<span class="k">for</span> <span class="n">fname</span><span class="p">,</span> <span class="n">fval</span> <span class="ow">in</span> <span class="n">featureset</span><span class="o">.</span><span class="n">items</span><span class="p">():</span>
<span class="n">positive_feature_freqdist</span><span class="p">[</span><span class="n">fname</span><span class="p">][</span><span class="n">fval</span><span class="p">]</span> <span class="o">+=</span> <span class="mi">1</span>
<span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">fval</span><span class="p">)</span>
<span class="n">fnames</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">fname</span><span class="p">)</span>
<span class="n">num_positive_examples</span> <span class="o">+=</span> <span class="mi">1</span>
<span class="c1"># Count up how many times each feature value occurred in unlabeled examples.</span>
<span class="n">num_unlabeled_examples</span> <span class="o">=</span> <span class="mi">0</span>
<span class="k">for</span> <span class="n">featureset</span> <span class="ow">in</span> <span class="n">unlabeled_featuresets</span><span class="p">:</span>
<span class="k">for</span> <span class="n">fname</span><span class="p">,</span> <span class="n">fval</span> <span class="ow">in</span> <span class="n">featureset</span><span class="o">.</span><span class="n">items</span><span class="p">():</span>
<span class="n">unlabeled_feature_freqdist</span><span class="p">[</span><span class="n">fname</span><span class="p">][</span><span class="n">fval</span><span class="p">]</span> <span class="o">+=</span> <span class="mi">1</span>
<span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">fval</span><span class="p">)</span>
<span class="n">fnames</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">fname</span><span class="p">)</span>
<span class="n">num_unlabeled_examples</span> <span class="o">+=</span> <span class="mi">1</span>
<span class="c1"># If a feature didn't have a value given for an instance, then we assume that</span>
<span class="c1"># it gets the implicit value 'None'.</span>
<span class="k">for</span> <span class="n">fname</span> <span class="ow">in</span> <span class="n">fnames</span><span class="p">:</span>
<span class="n">count</span> <span class="o">=</span> <span class="n">positive_feature_freqdist</span><span class="p">[</span><span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">N</span><span class="p">()</span>
<span class="n">positive_feature_freqdist</span><span class="p">[</span><span class="n">fname</span><span class="p">][</span><span class="kc">None</span><span class="p">]</span> <span class="o">+=</span> <span class="n">num_positive_examples</span> <span class="o">-</span> <span class="n">count</span>
<span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="kc">None</span><span class="p">)</span>
<span class="k">for</span> <span class="n">fname</span> <span class="ow">in</span> <span class="n">fnames</span><span class="p">:</span>
<span class="n">count</span> <span class="o">=</span> <span class="n">unlabeled_feature_freqdist</span><span class="p">[</span><span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">N</span><span class="p">()</span>
<span class="n">unlabeled_feature_freqdist</span><span class="p">[</span><span class="n">fname</span><span class="p">][</span><span class="kc">None</span><span class="p">]</span> <span class="o">+=</span> <span class="n">num_unlabeled_examples</span> <span class="o">-</span> <span class="n">count</span>
<span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="kc">None</span><span class="p">)</span>
<span class="n">negative_prob_prior</span> <span class="o">=</span> <span class="mf">1.0</span> <span class="o">-</span> <span class="n">positive_prob_prior</span>
<span class="c1"># Create the P(label) distribution.</span>
<span class="n">label_probdist</span> <span class="o">=</span> <span class="n">DictionaryProbDist</span><span class="p">(</span>
<span class="p">{</span><span class="kc">True</span><span class="p">:</span> <span class="n">positive_prob_prior</span><span class="p">,</span> <span class="kc">False</span><span class="p">:</span> <span class="n">negative_prob_prior</span><span class="p">}</span>
<span class="p">)</span>
<span class="c1"># Create the P(fval|label, fname) distribution.</span>
<span class="n">feature_probdist</span> <span class="o">=</span> <span class="p">{}</span>
<span class="k">for</span> <span class="n">fname</span><span class="p">,</span> <span class="n">freqdist</span> <span class="ow">in</span> <span class="n">positive_feature_freqdist</span><span class="o">.</span><span class="n">items</span><span class="p">():</span>
<span class="n">probdist</span> <span class="o">=</span> <span class="n">estimator</span><span class="p">(</span><span class="n">freqdist</span><span class="p">,</span> <span class="n">bins</span><span class="o">=</span><span class="nb">len</span><span class="p">(</span><span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]))</span>
<span class="n">feature_probdist</span><span class="p">[</span><span class="kc">True</span><span class="p">,</span> <span class="n">fname</span><span class="p">]</span> <span class="o">=</span> <span class="n">probdist</span>
<span class="k">for</span> <span class="n">fname</span><span class="p">,</span> <span class="n">freqdist</span> <span class="ow">in</span> <span class="n">unlabeled_feature_freqdist</span><span class="o">.</span><span class="n">items</span><span class="p">():</span>
<span class="n">global_probdist</span> <span class="o">=</span> <span class="n">estimator</span><span class="p">(</span><span class="n">freqdist</span><span class="p">,</span> <span class="n">bins</span><span class="o">=</span><span class="nb">len</span><span class="p">(</span><span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]))</span>
<span class="n">negative_feature_probs</span> <span class="o">=</span> <span class="p">{}</span>
<span class="k">for</span> <span class="n">fval</span> <span class="ow">in</span> <span class="n">feature_values</span><span class="p">[</span><span class="n">fname</span><span class="p">]:</span>
<span class="n">prob</span> <span class="o">=</span> <span class="p">(</span>
<span class="n">global_probdist</span><span class="o">.</span><span class="n">prob</span><span class="p">(</span><span class="n">fval</span><span class="p">)</span>
<span class="o">-</span> <span class="n">positive_prob_prior</span> <span class="o">*</span> <span class="n">feature_probdist</span><span class="p">[</span><span class="kc">True</span><span class="p">,</span> <span class="n">fname</span><span class="p">]</span><span class="o">.</span><span class="n">prob</span><span class="p">(</span><span class="n">fval</span><span class="p">)</span>
<span class="p">)</span> <span class="o">/</span> <span class="n">negative_prob_prior</span>
<span class="c1"># TODO: We need to add some kind of smoothing here, instead of</span>
<span class="c1"># setting negative probabilities to zero and normalizing.</span>
<span class="n">negative_feature_probs</span><span class="p">[</span><span class="n">fval</span><span class="p">]</span> <span class="o">=</span> <span class="nb">max</span><span class="p">(</span><span class="n">prob</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">)</span>
<span class="n">feature_probdist</span><span class="p">[</span><span class="kc">False</span><span class="p">,</span> <span class="n">fname</span><span class="p">]</span> <span class="o">=</span> <span class="n">DictionaryProbDist</span><span class="p">(</span>
<span class="n">negative_feature_probs</span><span class="p">,</span> <span class="n">normalize</span><span class="o">=</span><span class="kc">True</span>
<span class="p">)</span>
<span class="k">return</span> <span class="n">PositiveNaiveBayesClassifier</span><span class="p">(</span><span class="n">label_probdist</span><span class="p">,</span> <span class="n">feature_probdist</span><span class="p">)</span></div></div>
<span class="c1">##//////////////////////////////////////////////////////</span>
<span class="c1">## Demo</span>
<span class="c1">##//////////////////////////////////////////////////////</span>
<div class="viewcode-block" id="demo"><a class="viewcode-back" href="../../../api/nltk.classify.positivenaivebayes.html#nltk.classify.positivenaivebayes.demo">[docs]</a><span class="k">def</span> <span class="nf">demo</span><span class="p">():</span>
<span class="kn">from</span> <span class="nn">nltk.classify.util</span> <span class="kn">import</span> <span class="n">partial_names_demo</span>
<span class="n">classifier</span> <span class="o">=</span> <span class="n">partial_names_demo</span><span class="p">(</span><span class="n">PositiveNaiveBayesClassifier</span><span class="o">.</span><span class="n">train</span><span class="p">)</span>
<span class="n">classifier</span><span class="o">.</span><span class="n">show_most_informative_features</span><span class="p">()</span></div>
</pre></div>
</div>
</div>
</div>
<footer>
<div id="footer-info">
<ul id="build-details">
<li class="footer-element">
<a href="https://github.com/nltk/nltk/tree/3.6.5">3.6.5</a>
</li>
<li class="footer-element">
Oct 11, 2021
</li>
</ul>
<div id="copyright">
© 2021, NLTK Project
</div>
<div id="credit">
created with <a href="http://sphinx-doc.org/">Sphinx</a> and <a href="https://github.com/tomaarsen/nltk_theme">NLTK Theme</a>
</div>
</div>
</footer>
</div>
</body>
</html>