nltk 3.2__zip → 3.3__zip
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {nltk-3.2 → nltk-3.3}/INSTALL.txt +6 -6
- {nltk-3.2 → nltk-3.3}/LICENSE.txt +13 -13
- {nltk-3.2 → nltk-3.3}/MANIFEST.in +6 -6
- {nltk-3.2/nltk.egg-info → nltk-3.3}/PKG-INFO +13 -5
- nltk-3.2/README.txt → nltk-3.3/README.md +49 -48
- nltk-3.3/nltk/VERSION +1 -0
- {nltk-3.2 → nltk-3.3}/nltk/__init__.py +185 -185
- {nltk-3.2 → nltk-3.3}/nltk/app/__init__.py +52 -53
- {nltk-3.2 → nltk-3.3}/nltk/app/chartparser_app.py +2276 -2276
- {nltk-3.2 → nltk-3.3}/nltk/app/chunkparser_app.py +1262 -1264
- {nltk-3.2 → nltk-3.3}/nltk/app/collocations_app.py +347 -350
- {nltk-3.2 → nltk-3.3}/nltk/app/concordance_app.py +568 -570
- {nltk-3.2 → nltk-3.3}/nltk/app/nemo_app.py +156 -155
- {nltk-3.2 → nltk-3.3}/nltk/app/rdparser_app.py +892 -894
- {nltk-3.2 → nltk-3.3}/nltk/app/srparser_app.py +809 -809
- {nltk-3.2 → nltk-3.3}/nltk/app/wordfreq_app.py +32 -32
- {nltk-3.2 → nltk-3.3}/nltk/app/wordnet_app.py +968 -968
- {nltk-3.2 → nltk-3.3}/nltk/book.py +91 -90
- {nltk-3.2 → nltk-3.3}/nltk/ccg/__init__.py +22 -22
- {nltk-3.2 → nltk-3.3}/nltk/ccg/api.py +359 -335
- {nltk-3.2 → nltk-3.3}/nltk/ccg/chart.py +394 -361
- {nltk-3.2 → nltk-3.3}/nltk/ccg/combinator.py +343 -312
- {nltk-3.2 → nltk-3.3}/nltk/ccg/lexicon.py +328 -290
- nltk-3.3/nltk/ccg/logic.py +46 -0
- {nltk-3.2 → nltk-3.3}/nltk/chat/__init__.py +49 -49
- {nltk-3.2 → nltk-3.3}/nltk/chat/eliza.py +244 -244
- {nltk-3.2 → nltk-3.3}/nltk/chat/iesha.py +140 -140
- {nltk-3.2 → nltk-3.3}/nltk/chat/rude.py +92 -92
- {nltk-3.2 → nltk-3.3}/nltk/chat/suntsu.py +117 -117
- {nltk-3.2 → nltk-3.3}/nltk/chat/util.py +122 -120
- {nltk-3.2 → nltk-3.3}/nltk/chat/zen.py +282 -282
- {nltk-3.2 → nltk-3.3}/nltk/chunk/__init__.py +190 -190
- {nltk-3.2 → nltk-3.3}/nltk/chunk/api.py +51 -51
- {nltk-3.2 → nltk-3.3}/nltk/chunk/named_entity.py +331 -331
- {nltk-3.2 → nltk-3.3}/nltk/chunk/regexp.py +1390 -1384
- {nltk-3.2 → nltk-3.3}/nltk/chunk/util.py +600 -600
- {nltk-3.2 → nltk-3.3}/nltk/classify/__init__.py +98 -98
- {nltk-3.2 → nltk-3.3}/nltk/classify/api.py +193 -193
- {nltk-3.2 → nltk-3.3}/nltk/classify/decisiontree.py +295 -295
- {nltk-3.2 → nltk-3.3}/nltk/classify/maxent.py +1510 -1492
- {nltk-3.2 → nltk-3.3}/nltk/classify/megam.py +179 -178
- {nltk-3.2 → nltk-3.3}/nltk/classify/naivebayes.py +245 -242
- {nltk-3.2 → nltk-3.3}/nltk/classify/positivenaivebayes.py +170 -170
- {nltk-3.2 → nltk-3.3}/nltk/classify/rte_classify.py +160 -191
- {nltk-3.2 → nltk-3.3}/nltk/classify/scikitlearn.py +153 -151
- {nltk-3.2 → nltk-3.3}/nltk/classify/senna.py +183 -181
- {nltk-3.2 → nltk-3.3}/nltk/classify/svm.py +15 -15
- {nltk-3.2 → nltk-3.3}/nltk/classify/tadm.py +113 -112
- {nltk-3.2 → nltk-3.3}/nltk/classify/textcat.py +193 -193
- {nltk-3.2 → nltk-3.3}/nltk/classify/util.py +324 -312
- {nltk-3.2 → nltk-3.3}/nltk/classify/weka.py +346 -345
- {nltk-3.2 → nltk-3.3}/nltk/cluster/__init__.py +86 -86
- {nltk-3.2 → nltk-3.3}/nltk/cluster/api.py +74 -70
- {nltk-3.2 → nltk-3.3}/nltk/cluster/em.py +247 -247
- {nltk-3.2 → nltk-3.3}/nltk/cluster/gaac.py +168 -168
- {nltk-3.2 → nltk-3.3}/nltk/cluster/kmeans.py +221 -221
- {nltk-3.2 → nltk-3.3}/nltk/cluster/util.py +304 -289
- nltk-3.3/nltk/collections.py +640 -0
- {nltk-3.2 → nltk-3.3}/nltk/collocations.py +388 -388
- nltk-3.3/nltk/compat.py +365 -0
- {nltk-3.2 → nltk-3.3}/nltk/corpus/__init__.py +341 -320
- {nltk-3.2 → nltk-3.3}/nltk/corpus/europarl_raw.py +44 -44
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/__init__.py +147 -145
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/aligned.py +115 -114
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/api.py +448 -446
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/bnc.py +252 -252
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/bracket_parse.py +227 -227
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/categorized_sents.py +179 -178
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/chasen.py +140 -139
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/childes.py +522 -494
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/chunked.py +212 -212
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/cmudict.py +97 -95
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/comparative_sents.py +280 -278
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/conll.py +528 -521
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/crubadan.py +116 -116
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/dependency.py +101 -101
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/framenet.py +3059 -2093
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/ieer.py +112 -111
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/indian.py +85 -86
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/ipipan.py +332 -331
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/knbc.py +187 -187
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/lin.py +156 -156
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/mte.py +311 -310
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/nkjp.py +429 -428
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/nombank.py +421 -421
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/nps_chat.py +73 -73
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/opinion_lexicon.py +115 -115
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/panlex_lite.py +165 -165
- nltk-3.3/nltk/corpus/reader/pl196x.py +292 -0
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/plaintext.py +232 -233
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/ppattach.py +96 -95
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/propbank.py +479 -481
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/pros_cons.py +128 -126
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/reviews.py +330 -328
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/rte.py +144 -146
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/semcor.py +256 -256
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/senseval.py +203 -201
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/sentiwordnet.py +137 -137
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/sinica_treebank.py +75 -75
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/string_category.py +60 -59
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/switchboard.py +119 -119
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/tagged.py +295 -294
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/timit.py +452 -450
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/toolbox.py +68 -68
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/twitter.py +155 -157
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/udhr.py +80 -80
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/util.py +804 -800
- nltk-3.3/nltk/corpus/reader/verbnet.py +595 -0
- nltk-3.3/nltk/corpus/reader/wordlist.py +134 -0
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/wordnet.py +2108 -2036
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/xmldocs.py +392 -387
- {nltk-3.2 → nltk-3.3}/nltk/corpus/reader/ycoe.py +243 -242
- {nltk-3.2 → nltk-3.3}/nltk/corpus/util.py +144 -127
- {nltk-3.2 → nltk-3.3}/nltk/data.py +1487 -1451
- {nltk-3.2 → nltk-3.3}/nltk/decorators.py +219 -215
- {nltk-3.2 → nltk-3.3}/nltk/downloader.py +2278 -2274
- {nltk-3.2 → nltk-3.3}/nltk/draw/__init__.py +27 -27
- {nltk-3.2 → nltk-3.3}/nltk/draw/cfg.py +774 -777
- {nltk-3.2 → nltk-3.3}/nltk/draw/dispersion.py +58 -58
- {nltk-3.2 → nltk-3.3}/nltk/draw/table.py +1098 -1098
- {nltk-3.2 → nltk-3.3}/nltk/draw/tree.py +963 -966
- {nltk-3.2 → nltk-3.3}/nltk/draw/util.py +2356 -2356
- {nltk-3.2 → nltk-3.3}/nltk/featstruct.py +2617 -2500
- {nltk-3.2 → nltk-3.3}/nltk/grammar.py +1559 -1516
- {nltk-3.2 → nltk-3.3}/nltk/help.py +56 -56
- {nltk-3.2 → nltk-3.3}/nltk/inference/__init__.py +20 -20
- {nltk-3.2 → nltk-3.3}/nltk/inference/api.py +597 -589
- {nltk-3.2 → nltk-3.3}/nltk/inference/discourse.py +612 -608
- {nltk-3.2 → nltk-3.3}/nltk/inference/mace.py +311 -311
- {nltk-3.2 → nltk-3.3}/nltk/inference/nonmonotonic.py +509 -509
- {nltk-3.2 → nltk-3.3}/nltk/inference/prover9.py +431 -431
- {nltk-3.2 → nltk-3.3}/nltk/inference/resolution.py +687 -687
- {nltk-3.2 → nltk-3.3}/nltk/inference/tableau.py +607 -607
- {nltk-3.2 → nltk-3.3}/nltk/internals.py +982 -982
- {nltk-3.2 → nltk-3.3}/nltk/jsontags.py +63 -63
- {nltk-3.2 → nltk-3.3}/nltk/lazyimport.py +142 -142
- {nltk-3.2 → nltk-3.3}/nltk/metrics/__init__.py +30 -29
- {nltk-3.2 → nltk-3.3}/nltk/metrics/agreement.py +421 -401
- nltk-3.3/nltk/metrics/aline.py +607 -0
- {nltk-3.2 → nltk-3.3}/nltk/metrics/association.py +414 -410
- {nltk-3.2 → nltk-3.3}/nltk/metrics/confusionmatrix.py +206 -206
- {nltk-3.2 → nltk-3.3}/nltk/metrics/distance.py +207 -202
- {nltk-3.2 → nltk-3.3}/nltk/metrics/paice.py +381 -381
- {nltk-3.2 → nltk-3.3}/nltk/metrics/scores.py +228 -228
- {nltk-3.2 → nltk-3.3}/nltk/metrics/segmentation.py +231 -232
- {nltk-3.2 → nltk-3.3}/nltk/metrics/spearman.py +69 -69
- {nltk-3.2 → nltk-3.3}/nltk/misc/__init__.py +11 -11
- {nltk-3.2 → nltk-3.3}/nltk/misc/babelfish.py +10 -10
- {nltk-3.2 → nltk-3.3}/nltk/misc/chomsky.py +135 -133
- {nltk-3.2 → nltk-3.3}/nltk/misc/minimalset.py +83 -83
- {nltk-3.2 → nltk-3.3}/nltk/misc/sort.py +157 -157
- {nltk-3.2 → nltk-3.3}/nltk/misc/wordfinder.py +129 -129
- {nltk-3.2 → nltk-3.3}/nltk/parse/__init__.py +81 -80
- {nltk-3.2 → nltk-3.3}/nltk/parse/api.py +66 -66
- {nltk-3.2 → nltk-3.3}/nltk/parse/bllip.py +282 -282
- {nltk-3.2 → nltk-3.3}/nltk/parse/chart.py +1682 -1681
- nltk-3.3/nltk/parse/corenlp.py +780 -0
- {nltk-3.2 → nltk-3.3}/nltk/parse/dependencygraph.py +760 -758
- {nltk-3.2 → nltk-3.3}/nltk/parse/earleychart.py +452 -451
- {nltk-3.2 → nltk-3.3}/nltk/parse/evaluate.py +129 -129
- {nltk-3.2 → nltk-3.3}/nltk/parse/featurechart.py +580 -578
- {nltk-3.2 → nltk-3.3}/nltk/parse/generate.py +87 -76
- {nltk-3.2 → nltk-3.3}/nltk/parse/malt.py +356 -356
- {nltk-3.2 → nltk-3.3}/nltk/parse/nonprojectivedependencyparser.py +783 -783
- {nltk-3.2 → nltk-3.3}/nltk/parse/pchart.py +522 -522
- {nltk-3.2 → nltk-3.3}/nltk/parse/projectivedependencyparser.py +579 -578
- {nltk-3.2 → nltk-3.3}/nltk/parse/recursivedescent.py +655 -656
- {nltk-3.2 → nltk-3.3}/nltk/parse/shiftreduce.py +458 -459
- {nltk-3.2 → nltk-3.3}/nltk/parse/stanford.py +431 -406
- {nltk-3.2 → nltk-3.3}/nltk/parse/transitionparser.py +774 -772
- {nltk-3.2 → nltk-3.3}/nltk/parse/util.py +231 -231
- {nltk-3.2 → nltk-3.3}/nltk/parse/viterbi.py +401 -401
- {nltk-3.2 → nltk-3.3}/nltk/probability.py +2410 -2355
- {nltk-3.2 → nltk-3.3}/nltk/sem/__init__.py +61 -61
- {nltk-3.2 → nltk-3.3}/nltk/sem/boxer.py +1261 -1261
- {nltk-3.2 → nltk-3.3}/nltk/sem/chat80.py +782 -782
- {nltk-3.2 → nltk-3.3}/nltk/sem/cooper_storage.py +118 -118
- {nltk-3.2 → nltk-3.3}/nltk/sem/drt.py +1258 -1257
- {nltk-3.2 → nltk-3.3}/nltk/sem/drt_glue_demo.py +485 -489
- {nltk-3.2 → nltk-3.3}/nltk/sem/evaluate.py +790 -784
- {nltk-3.2 → nltk-3.3}/nltk/sem/glue.py +696 -694
- {nltk-3.2 → nltk-3.3}/nltk/sem/hole.py +386 -384
- {nltk-3.2 → nltk-3.3}/nltk/sem/lfg.py +210 -208
- {nltk-3.2 → nltk-3.3}/nltk/sem/linearlogic.py +451 -449
- {nltk-3.2 → nltk-3.3}/nltk/sem/logic.py +1910 -1910
- {nltk-3.2 → nltk-3.3}/nltk/sem/relextract.py +476 -483
- {nltk-3.2 → nltk-3.3}/nltk/sem/skolemize.py +101 -101
- {nltk-3.2 → nltk-3.3}/nltk/sem/util.py +249 -249
- {nltk-3.2 → nltk-3.3}/nltk/sentiment/__init__.py +14 -14
- {nltk-3.2 → nltk-3.3}/nltk/sentiment/sentiment_analyzer.py +229 -229
- {nltk-3.2 → nltk-3.3}/nltk/sentiment/util.py +762 -752
- {nltk-3.2 → nltk-3.3}/nltk/sentiment/vader.py +443 -445
- {nltk-3.2 → nltk-3.3}/nltk/stem/__init__.py +31 -33
- {nltk-3.2 → nltk-3.3}/nltk/stem/api.py +28 -25
- nltk-3.3/nltk/stem/arlstem.py +369 -0
- {nltk-3.2 → nltk-3.3}/nltk/stem/isri.py +345 -345
- {nltk-3.2 → nltk-3.3}/nltk/stem/lancaster.py +337 -310
- nltk-3.3/nltk/stem/porter.py +710 -0
- {nltk-3.2 → nltk-3.3}/nltk/stem/regexp.py +61 -61
- {nltk-3.2 → nltk-3.3}/nltk/stem/rslp.py +142 -142
- {nltk-3.2 → nltk-3.3}/nltk/stem/snowball.py +4236 -3712
- {nltk-3.2 → nltk-3.3}/nltk/stem/util.py +22 -12
- {nltk-3.2 → nltk-3.3}/nltk/stem/wordnet.py +51 -51
- {nltk-3.2 → nltk-3.3}/nltk/tag/__init__.py +152 -127
- {nltk-3.2 → nltk-3.3}/nltk/tag/api.py +86 -81
- {nltk-3.2 → nltk-3.3}/nltk/tag/brill.py +424 -426
- {nltk-3.2 → nltk-3.3}/nltk/tag/brill_trainer.py +608 -608
- {nltk-3.2 → nltk-3.3}/nltk/tag/crf.py +204 -200
- {nltk-3.2 → nltk-3.3}/nltk/tag/hmm.py +1277 -1277
- {nltk-3.2 → nltk-3.3}/nltk/tag/hunpos.py +132 -132
- {nltk-3.2 → nltk-3.3}/nltk/tag/mapping.py +101 -101
- {nltk-3.2 → nltk-3.3}/nltk/tag/perceptron.py +329 -314
- {nltk-3.2 → nltk-3.3}/nltk/tag/senna.py +147 -145
- {nltk-3.2 → nltk-3.3}/nltk/tag/sequential.py +740 -736
- {nltk-3.2 → nltk-3.3}/nltk/tag/stanford.py +219 -196
- {nltk-3.2 → nltk-3.3}/nltk/tag/tnt.py +595 -604
- {nltk-3.2 → nltk-3.3}/nltk/tag/util.py +72 -72
- {nltk-3.2 → nltk-3.3}/nltk/tbl/__init__.py +28 -28
- {nltk-3.2 → nltk-3.3}/nltk/tbl/api.py +1 -1
- {nltk-3.2 → nltk-3.3}/nltk/tbl/demo.py +366 -366
- {nltk-3.2 → nltk-3.3}/nltk/tbl/erroranalysis.py +39 -39
- {nltk-3.2 → nltk-3.3}/nltk/tbl/feature.py +259 -261
- {nltk-3.2 → nltk-3.3}/nltk/tbl/rule.py +321 -321
- {nltk-3.2 → nltk-3.3}/nltk/tbl/template.py +303 -308
- {nltk-3.2 → nltk-3.3}/nltk/test/__init__.py +18 -18
- {nltk-3.2 → nltk-3.3}/nltk/test/all.py +24 -23
- {nltk-3.2 → nltk-3.3}/nltk/test/bleu.doctest +14 -14
- {nltk-3.2 → nltk-3.3}/nltk/test/bnc.doctest +60 -60
- {nltk-3.2 → nltk-3.3}/nltk/test/ccg.doctest +376 -395
- nltk-3.3/nltk/test/ccg_semantics.doctest +553 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/chat80.doctest +234 -234
- {nltk-3.2 → nltk-3.3}/nltk/test/childes.doctest +184 -184
- {nltk-3.2 → nltk-3.3}/nltk/test/childes_fixt.py +14 -13
- {nltk-3.2 → nltk-3.3}/nltk/test/chunk.doctest +373 -373
- {nltk-3.2 → nltk-3.3}/nltk/test/classify.doctest +185 -185
- {nltk-3.2 → nltk-3.3}/nltk/test/classify_fixt.py +11 -10
- nltk-3.3/nltk/test/collections.doctest +20 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/collocations.doctest +276 -276
- {nltk-3.2 → nltk-3.3}/nltk/test/compat.doctest +134 -138
- {nltk-3.2 → nltk-3.3}/nltk/test/compat_fixt.py +9 -8
- nltk-3.3/nltk/test/concordance.doctest +55 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/corpus.doctest +2200 -2212
- {nltk-3.2 → nltk-3.3}/nltk/test/corpus_fixt.py +3 -3
- {nltk-3.2 → nltk-3.3}/nltk/test/crubadan.doctest +65 -65
- {nltk-3.2 → nltk-3.3}/nltk/test/data.doctest +379 -379
- {nltk-3.2 → nltk-3.3}/nltk/test/dependency.doctest +241 -241
- {nltk-3.2 → nltk-3.3}/nltk/test/discourse.doctest +546 -546
- {nltk-3.2 → nltk-3.3}/nltk/test/discourse_fixt.py +14 -13
- {nltk-3.2 → nltk-3.3}/nltk/test/doctest_nose_plugin.py +159 -156
- {nltk-3.2 → nltk-3.3}/nltk/test/drt.doctest +517 -517
- {nltk-3.2 → nltk-3.3}/nltk/test/featgram.doctest +607 -607
- {nltk-3.2 → nltk-3.3}/nltk/test/featstruct.doctest +1229 -1230
- {nltk-3.2 → nltk-3.3}/nltk/test/framenet.doctest +288 -239
- {nltk-3.2 → nltk-3.3}/nltk/test/generate.doctest +67 -67
- {nltk-3.2 → nltk-3.3}/nltk/test/gensim.doctest +140 -141
- {nltk-3.2 → nltk-3.3}/nltk/test/gensim_fixt.py +10 -9
- {nltk-3.2 → nltk-3.3}/nltk/test/gluesemantics.doctest +384 -386
- {nltk-3.2 → nltk-3.3}/nltk/test/gluesemantics_malt.doctest +68 -68
- {nltk-3.2 → nltk-3.3}/nltk/test/gluesemantics_malt_fixt.py +12 -11
- {nltk-3.2 → nltk-3.3}/nltk/test/grammar.doctest +48 -48
- {nltk-3.2 → nltk-3.3}/nltk/test/grammartestsuites.doctest +109 -109
- {nltk-3.2 → nltk-3.3}/nltk/test/index.doctest +100 -100
- {nltk-3.2 → nltk-3.3}/nltk/test/inference.doctest +534 -534
- {nltk-3.2 → nltk-3.3}/nltk/test/inference_fixt.py +12 -11
- {nltk-3.2 → nltk-3.3}/nltk/test/internals.doctest +140 -140
- {nltk-3.2 → nltk-3.3}/nltk/test/japanese.doctest +48 -48
- {nltk-3.2 → nltk-3.3}/nltk/test/logic.doctest +1098 -1098
- {nltk-3.2 → nltk-3.3}/nltk/test/metrics.doctest +270 -270
- {nltk-3.2 → nltk-3.3}/nltk/test/misc.doctest +118 -118
- {nltk-3.2 → nltk-3.3}/nltk/test/nonmonotonic.doctest +286 -286
- {nltk-3.2 → nltk-3.3}/nltk/test/nonmonotonic_fixt.py +12 -11
- {nltk-3.2 → nltk-3.3}/nltk/test/paice.doctest +35 -35
- {nltk-3.2 → nltk-3.3}/nltk/test/parse.doctest +884 -884
- {nltk-3.2 → nltk-3.3}/nltk/test/portuguese_en.doctest +565 -565
- {nltk-3.2 → nltk-3.3}/nltk/test/portuguese_en_fixt.py +16 -16
- {nltk-3.2 → nltk-3.3}/nltk/test/probability.doctest +304 -304
- {nltk-3.2 → nltk-3.3}/nltk/test/probability_fixt.py +13 -12
- {nltk-3.2 → nltk-3.3}/nltk/test/propbank.doctest +176 -176
- {nltk-3.2 → nltk-3.3}/nltk/test/relextract.doctest +263 -263
- {nltk-3.2 → nltk-3.3}/nltk/test/resolution.doctest +221 -221
- {nltk-3.2 → nltk-3.3}/nltk/test/runtests.py +86 -64
- {nltk-3.2 → nltk-3.3}/nltk/test/segmentation_fixt.py +10 -10
- {nltk-3.2 → nltk-3.3}/nltk/test/semantics.doctest +665 -665
- {nltk-3.2 → nltk-3.3}/nltk/test/semantics_fixt.py +7 -7
- {nltk-3.2 → nltk-3.3}/nltk/test/sentiment.doctest +233 -233
- {nltk-3.2 → nltk-3.3}/nltk/test/sentiwordnet.doctest +41 -41
- {nltk-3.2 → nltk-3.3}/nltk/test/simple.doctest +85 -85
- {nltk-3.2 → nltk-3.3}/nltk/test/stem.doctest +78 -78
- {nltk-3.2 → nltk-3.3}/nltk/test/tag.doctest +33 -33
- {nltk-3.2 → nltk-3.3}/nltk/test/tokenize.doctest +273 -166
- {nltk-3.2 → nltk-3.3}/nltk/test/toolbox.doctest +307 -307
- {nltk-3.2 → nltk-3.3}/nltk/test/translate.doctest +242 -242
- {nltk-3.2 → nltk-3.3}/nltk/test/translate_fixt.py +3 -3
- {nltk-3.2 → nltk-3.3}/nltk/test/tree.doctest +1101 -1101
- {nltk-3.2 → nltk-3.3}/nltk/test/treeprettyprinter.doctest +127 -127
- {nltk-3.2 → nltk-3.3}/nltk/test/treetransforms.doctest +156 -156
- nltk-3.3/nltk/test/unit/test_2x_compat.py +63 -0
- nltk-3.3/nltk/test/unit/test_aline.py +52 -0
- nltk-3.3/nltk/test/unit/test_chunk.py +56 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_classify.py +55 -53
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_collocations.py +88 -79
- nltk-3.3/nltk/test/unit/test_concordance.py +98 -0
- nltk-3.3/nltk/test/unit/test_corenlp.py +1000 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_corpora.py +200 -183
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_corpus_views.py +45 -45
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_hmm.py +88 -88
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_json2csv_corpus.py +188 -187
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_naivebayes.py +26 -26
- nltk-3.3/nltk/test/unit/test_rte_classify.py +86 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_seekable_unicode_stream_reader.py +124 -124
- nltk-3.3/nltk/test/unit/test_senna.py +74 -0
- nltk-3.3/nltk/test/unit/test_stem.py +121 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_tag.py +20 -20
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_tgrep.py +628 -626
- nltk-3.3/nltk/test/unit/test_tokenize.py +168 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/test_twitter_auth.py +185 -185
- nltk-3.3/nltk/test/unit/test_wordnet.py +134 -0
- nltk-3.3/nltk/test/unit/translate/test_bleu.py +244 -0
- nltk-3.3/nltk/test/unit/translate/test_gdfa.py +57 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_ibm1.py +75 -75
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_ibm2.py +87 -87
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_ibm3.py +105 -105
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_ibm4.py +126 -126
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_ibm5.py +166 -166
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_ibm_model.py +270 -270
- nltk-3.3/nltk/test/unit/translate/test_nist.py +37 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/test_stack_decoder.py +300 -300
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/utils.py +41 -41
- {nltk-3.2 → nltk-3.3}/nltk/test/util.doctest +48 -48
- {nltk-3.2 → nltk-3.3}/nltk/test/wordnet.doctest +595 -581
- {nltk-3.2 → nltk-3.3}/nltk/test/wordnet_fixt.py +7 -6
- {nltk-3.2 → nltk-3.3}/nltk/test/wordnet_lch.doctest +53 -53
- {nltk-3.2 → nltk-3.3}/nltk/test/wsd.doctest +68 -68
- {nltk-3.2 → nltk-3.3}/nltk/text.py +663 -610
- {nltk-3.2 → nltk-3.3}/nltk/tgrep.py +938 -937
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/__init__.py +130 -107
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/api.py +77 -78
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/casual.py +343 -342
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/mwe.py +120 -120
- nltk-3.3/nltk/tokenize/nist.py +173 -0
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/punkt.py +1620 -1601
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/regexp.py +208 -202
- nltk-3.3/nltk/tokenize/repp.py +151 -0
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/sexpr.py +140 -142
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/simple.py +138 -136
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/stanford.py +117 -113
- nltk-3.3/nltk/tokenize/stanford_segmenter.py +246 -0
- {nltk-3.2 → nltk-3.3}/nltk/tokenize/texttiling.py +458 -460
- nltk-3.3/nltk/tokenize/toktok.py +155 -0
- nltk-3.3/nltk/tokenize/treebank.py +363 -0
- nltk-3.3/nltk/tokenize/util.py +261 -0
- {nltk-3.2 → nltk-3.3}/nltk/toolbox.py +499 -497
- {nltk-3.2 → nltk-3.3}/nltk/translate/__init__.py +24 -24
- {nltk-3.2 → nltk-3.3}/nltk/translate/api.py +331 -321
- {nltk-3.2 → nltk-3.3}/nltk/translate/bleu_score.py +615 -556
- nltk-3.3/nltk/translate/chrf_score.py +137 -0
- {nltk-3.2 → nltk-3.3}/nltk/translate/gale_church.py +248 -235
- {nltk-3.2 → nltk-3.3}/nltk/translate/gdfa.py +135 -131
- nltk-3.3/nltk/translate/gleu_score.py +193 -0
- {nltk-3.2 → nltk-3.3}/nltk/translate/ibm1.py +250 -241
- {nltk-3.2 → nltk-3.3}/nltk/translate/ibm2.py +308 -308
- {nltk-3.2 → nltk-3.3}/nltk/translate/ibm3.py +337 -337
- {nltk-3.2 → nltk-3.3}/nltk/translate/ibm4.py +467 -467
- {nltk-3.2 → nltk-3.3}/nltk/translate/ibm5.py +639 -639
- {nltk-3.2 → nltk-3.3}/nltk/translate/ibm_model.py +536 -533
- {nltk-3.2 → nltk-3.3}/nltk/translate/metrics.py +40 -40
- nltk-3.3/nltk/translate/nist_score.py +178 -0
- {nltk-3.2 → nltk-3.3}/nltk/translate/phrase_based.py +175 -175
- {nltk-3.2 → nltk-3.3}/nltk/translate/ribes_score.py +325 -325
- {nltk-3.2 → nltk-3.3}/nltk/translate/stack_decoder.py +499 -495
- {nltk-3.2 → nltk-3.3}/nltk/tree.py +1605 -1603
- {nltk-3.2 → nltk-3.3}/nltk/treeprettyprinter.py +564 -564
- {nltk-3.2 → nltk-3.3}/nltk/treetransforms.py +309 -309
- {nltk-3.2 → nltk-3.3}/nltk/twitter/__init__.py +28 -28
- {nltk-3.2 → nltk-3.3}/nltk/twitter/api.py +141 -136
- {nltk-3.2 → nltk-3.3}/nltk/twitter/common.py +264 -258
- {nltk-3.2 → nltk-3.3}/nltk/twitter/twitter_demo.py +288 -288
- {nltk-3.2 → nltk-3.3}/nltk/twitter/twitterclient.py +530 -530
- {nltk-3.2 → nltk-3.3}/nltk/twitter/util.py +147 -146
- nltk-3.3/nltk/util.py +751 -0
- {nltk-3.2 → nltk-3.3}/nltk/wsd.py +53 -53
- {nltk-3.2 → nltk-3.3/nltk.egg-info}/PKG-INFO +13 -5
- {nltk-3.2 → nltk-3.3}/nltk.egg-info/SOURCES.txt +26 -3
- nltk-3.3/nltk.egg-info/requires.txt +31 -0
- {nltk-3.2 → nltk-3.3}/setup.cfg +0 -1
- {nltk-3.2 → nltk-3.3}/setup.py +102 -75
- nltk-3.2/nltk/VERSION +0 -1
- nltk-3.2/nltk/compat.py +0 -680
- nltk-3.2/nltk/corpus/reader/pl196x.py +0 -278
- nltk-3.2/nltk/corpus/reader/verbnet.py +0 -408
- nltk-3.2/nltk/corpus/reader/wordlist.py +0 -37
- nltk-3.2/nltk/sentiment/vader_lexicon.txt +0 -7517
- nltk-3.2/nltk/six.py +0 -868
- nltk-3.2/nltk/stem/porter.py +0 -692
- nltk-3.2/nltk/test/unit/test_2x_compat.py +0 -26
- nltk-3.2/nltk/test/unit/test_stem.py +0 -36
- nltk-3.2/nltk/test/unit/test_tokenize.py +0 -23
- nltk-3.2/nltk/test/unit/translate/test_bleu.py +0 -84
- nltk-3.2/nltk/tokenize/treebank.py +0 -124
- nltk-3.2/nltk/tokenize/util.py +0 -90
- nltk-3.2/nltk/util.py +0 -1364
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/__init__.py +0 -0
- {nltk-3.2 → nltk-3.3}/nltk/test/unit/translate/__init__.py +0 -0
- {nltk-3.2 → nltk-3.3}/nltk.egg-info/dependency_links.txt +0 -0
- {nltk-3.2 → nltk-3.3}/nltk.egg-info/not-zip-safe +0 -0
- {nltk-3.2 → nltk-3.3}/nltk.egg-info/top_level.txt +0 -0
|
@@ -1,565 +1,565 @@
|
|
|
1
|
-
.. Copyright (C) 2001-
|
|
2
|
-
.. For license information, see LICENSE.TXT
|
|
3
|
-
|
|
4
|
-
==================================
|
|
5
|
-
Examples for Portuguese Processing
|
|
6
|
-
==================================
|
|
7
|
-
|
|
8
|
-
This HOWTO contains a variety of examples relating to the Portuguese language.
|
|
9
|
-
It is intended to be read in conjunction with the NLTK book
|
|
10
|
-
(``http://nltk.org/book``). For instructions on running the Python
|
|
11
|
-
interpreter, please see the section *Getting Started with Python*, in Chapter 1.
|
|
12
|
-
|
|
13
|
-
--------------------------------------------
|
|
14
|
-
Python Programming, with Portuguese Examples
|
|
15
|
-
--------------------------------------------
|
|
16
|
-
|
|
17
|
-
Chapter 1 of the NLTK book contains many elementary programming examples, all
|
|
18
|
-
with English texts. In this section, we'll see some corresponding examples
|
|
19
|
-
using Portuguese. Please refer to the chapter for full discussion. *Vamos!*
|
|
20
|
-
|
|
21
|
-
>>> from nltk.examples.pt import *
|
|
22
|
-
*** Introductory Examples for the NLTK Book ***
|
|
23
|
-
Loading ptext1, ... and psent1, ...
|
|
24
|
-
Type the name of the text or sentence to view it.
|
|
25
|
-
Type: 'texts()' or 'sents()' to list the materials.
|
|
26
|
-
ptext1: Memórias Póstumas de Brás Cubas (1881)
|
|
27
|
-
ptext2: Dom Casmurro (1899)
|
|
28
|
-
ptext3: Gênesis
|
|
29
|
-
ptext4: Folha de Sao Paulo (1994)
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
Any time we want to find out about these texts, we just have
|
|
33
|
-
to enter their names at the Python prompt:
|
|
34
|
-
|
|
35
|
-
>>> ptext2
|
|
36
|
-
<Text: Dom Casmurro (1899)>
|
|
37
|
-
|
|
38
|
-
Searching Text
|
|
39
|
-
--------------
|
|
40
|
-
|
|
41
|
-
A concordance permits us to see words in context.
|
|
42
|
-
|
|
43
|
-
>>> ptext1.concordance('olhos')
|
|
44
|
-
Building index...
|
|
45
|
-
Displaying 25 of 138 matches:
|
|
46
|
-
De pé , à cabeceira da cama , com os olhos estúpidos , a boca entreaberta , a t
|
|
47
|
-
orelhas . Pela minha parte fechei os olhos e deixei - me ir à ventura . Já agor
|
|
48
|
-
xões de cérebro enfermo . Como ia de olhos fechados , não via o caminho ; lembr
|
|
49
|
-
gelos eternos . Com efeito , abri os olhos e vi que o meu animal galopava numa
|
|
50
|
-
me apareceu então , fitando - me uns olhos rutilantes como o sol . Tudo nessa f
|
|
51
|
-
mim mesmo . Então , encarei - a com olhos súplices , e pedi mais alguns anos .
|
|
52
|
-
...
|
|
53
|
-
|
|
54
|
-
For a given word, we can find words with a similar text distribution:
|
|
55
|
-
|
|
56
|
-
>>> ptext1.similar('chegar')
|
|
57
|
-
Building word-context index...
|
|
58
|
-
acabada acudir aludir avistar bramanismo casamento cheguei com contar
|
|
59
|
-
contrário corpo dali deixei desferirem dizer fazer filhos já leitor lhe
|
|
60
|
-
>>> ptext3.similar('chegar')
|
|
61
|
-
Building word-context index...
|
|
62
|
-
achar alumiar arrombar destruir governar guardar ir lavrar passar que
|
|
63
|
-
toda tomar ver vir
|
|
64
|
-
|
|
65
|
-
We can search for the statistically significant collocations in a text:
|
|
66
|
-
|
|
67
|
-
>>> ptext1.collocations()
|
|
68
|
-
Building collocations list
|
|
69
|
-
Quincas Borba; Lobo Neves; alguma coisa; Brás Cubas; meu pai; dia
|
|
70
|
-
seguinte; não sei; Meu pai; alguns instantes; outra vez; outra coisa;
|
|
71
|
-
por exemplo; mim mesmo; coisa nenhuma; mesma coisa; não era; dias
|
|
72
|
-
depois; Passeio Público; olhar para; das coisas
|
|
73
|
-
|
|
74
|
-
We can search for words in context, with the help of *regular expressions*, e.g.:
|
|
75
|
-
|
|
76
|
-
>>> ptext1.findall("<olhos> (<.*>)")
|
|
77
|
-
estúpidos; e; fechados; rutilantes; súplices; a; do; babavam;
|
|
78
|
-
na; moles; se; da; umas; espraiavam; chamejantes; espetados;
|
|
79
|
-
...
|
|
80
|
-
|
|
81
|
-
We can automatically generate random text based on a given text, e.g.:
|
|
82
|
-
|
|
83
|
-
>>> ptext3.generate() # doctest: +SKIP
|
|
84
|
-
No princípio , criou Deus os abençoou , dizendo : Onde { estão } e até
|
|
85
|
-
à ave dos céus , { que } será . Disse mais Abrão : Dá - me a mulher
|
|
86
|
-
que tomaste ; porque daquele poço Eseque , { tinha .} E disse : Não
|
|
87
|
-
poderemos descer ; mas , do campo ainda não estava na casa do teu
|
|
88
|
-
pescoço . E viveu Serugue , depois Simeão e Levi { são } estes ? E o
|
|
89
|
-
varão , porque habitava na terra de Node , da mão de Esaú : Jeús ,
|
|
90
|
-
Jalão e Corá
|
|
91
|
-
|
|
92
|
-
Texts as List of Words
|
|
93
|
-
----------------------
|
|
94
|
-
|
|
95
|
-
A few sentences have been defined for you.
|
|
96
|
-
|
|
97
|
-
>>> psent1
|
|
98
|
-
['o', 'amor', 'da', 'gl\xf3ria', 'era', 'a', 'coisa', 'mais',
|
|
99
|
-
'verdadeiramente', 'humana', 'que', 'h\xe1', 'no', 'homem', ',',
|
|
100
|
-
'e', ',', 'conseq\xfcentemente', ',', 'a', 'sua', 'mais',
|
|
101
|
-
'genu\xedna', 'fei\xe7\xe3o', '.']
|
|
102
|
-
>>>
|
|
103
|
-
|
|
104
|
-
Notice that the sentence has been *tokenized*. Each token is
|
|
105
|
-
represented as a string, represented using quotes, e.g. ``'coisa'``.
|
|
106
|
-
Some strings contain special characters, e.g. ``\xf3``,
|
|
107
|
-
the internal representation for ó.
|
|
108
|
-
The tokens are combined in the form of a *list*. How long is this list?
|
|
109
|
-
|
|
110
|
-
>>> len(psent1)
|
|
111
|
-
25
|
|
112
|
-
>>>
|
|
113
|
-
|
|
114
|
-
What is the vocabulary of this sentence?
|
|
115
|
-
|
|
116
|
-
>>> sorted(set(psent1))
|
|
117
|
-
[',', '.', 'a', 'amor', 'coisa', 'conseqüentemente', 'da', 'e', 'era',
|
|
118
|
-
'feição', 'genuína', 'glória', 'homem', 'humana', 'há', 'mais', 'no',
|
|
119
|
-
'o', 'que', 'sua', 'verdadeiramente']
|
|
120
|
-
>>>
|
|
121
|
-
|
|
122
|
-
Let's iterate over each item in ``psent2``, and print information for each:
|
|
123
|
-
|
|
124
|
-
>>> for w in psent2:
|
|
125
|
-
... print(w, len(w), w[-1])
|
|
126
|
-
...
|
|
127
|
-
Não 3 o
|
|
128
|
-
consultes 9 s
|
|
129
|
-
dicionários 11 s
|
|
130
|
-
. 1 .
|
|
131
|
-
|
|
132
|
-
Observe how we make a human-readable version of a string, using ``decode()``.
|
|
133
|
-
Also notice that we accessed the last character of a string ``w`` using ``w[-1]``.
|
|
134
|
-
|
|
135
|
-
We just saw a ``for`` loop above. Another useful control structure is a
|
|
136
|
-
*list comprehension*.
|
|
137
|
-
|
|
138
|
-
>>> [w.upper() for w in psent2]
|
|
139
|
-
['N\xc3O', 'CONSULTES', 'DICION\xc1RIOS', '.']
|
|
140
|
-
>>> [w for w in psent1 if w.endswith('a')]
|
|
141
|
-
['da', 'gl\xf3ria', 'era', 'a', 'coisa', 'humana', 'a', 'sua', 'genu\xedna']
|
|
142
|
-
>>> [w for w in ptext4 if len(w) > 15]
|
|
143
|
-
[u'norte-irlandeses', u'pan-nacionalismo', u'predominatemente', u'primeiro-ministro',
|
|
144
|
-
u'primeiro-ministro', u'irlandesa-americana', u'responsabilidades', u'significativamente']
|
|
145
|
-
|
|
146
|
-
We can examine the relative frequency of words in a text, using ``FreqDist``:
|
|
147
|
-
|
|
148
|
-
>>> fd1 = FreqDist(ptext1)
|
|
149
|
-
>>> fd1
|
|
150
|
-
<FreqDist with 10848 samples and 77098 outcomes>
|
|
151
|
-
>>> fd1['olhos']
|
|
152
|
-
137
|
|
153
|
-
>>> fd1.max()
|
|
154
|
-
u','
|
|
155
|
-
>>> fd1.samples()[:100]
|
|
156
|
-
[u',', u'.', u'a', u'que', u'de', u'e', u'-', u'o', u';', u'me', u'um', u'n\xe3o',
|
|
157
|
-
u'\x97', u'se', u'do', u'da', u'uma', u'com', u'os', u'\xe9', u'era', u'as', u'eu',
|
|
158
|
-
u'lhe', u'ao', u'em', u'para', u'mas', u'...', u'!', u'\xe0', u'na', u'mais', u'?',
|
|
159
|
-
u'no', u'como', u'por', u'N\xe3o', u'dos', u'ou', u'ele', u':', u'Virg\xedlia',
|
|
160
|
-
u'meu', u'disse', u'minha', u'das', u'O', u'/', u'A', u'CAP\xcdTULO', u'muito',
|
|
161
|
-
u'depois', u'coisa', u'foi', u'sem', u'olhos', u'ela', u'nos', u'tinha', u'nem',
|
|
162
|
-
u'E', u'outro', u'vida', u'nada', u'tempo', u'menos', u'outra', u'casa', u'homem',
|
|
163
|
-
u'porque', u'quando', u'mim', u'mesmo', u'ser', u'pouco', u'estava', u'dia',
|
|
164
|
-
u't\xe3o', u'tudo', u'Mas', u'at\xe9', u'D', u'ainda', u's\xf3', u'alguma',
|
|
165
|
-
u'la', u'vez', u'anos', u'h\xe1', u'Era', u'pai', u'esse', u'lo', u'dizer', u'assim',
|
|
166
|
-
u'ent\xe3o', u'dizia', u'aos', u'Borba']
|
|
167
|
-
|
|
168
|
-
---------------
|
|
169
|
-
Reading Corpora
|
|
170
|
-
---------------
|
|
171
|
-
|
|
172
|
-
Accessing the Machado Text Corpus
|
|
173
|
-
---------------------------------
|
|
174
|
-
|
|
175
|
-
NLTK includes the complete works of Machado de Assis.
|
|
176
|
-
|
|
177
|
-
>>> from nltk.corpus import machado
|
|
178
|
-
>>> machado.fileids()
|
|
179
|
-
['contos/macn001.txt', 'contos/macn002.txt', 'contos/macn003.txt', ...]
|
|
180
|
-
|
|
181
|
-
Each file corresponds to one of the works of Machado de Assis. To see a complete
|
|
182
|
-
list of works, you can look at the corpus README file: ``print machado.readme()``.
|
|
183
|
-
Let's access the text of the *Posthumous Memories of Brás Cubas*.
|
|
184
|
-
|
|
185
|
-
We can access the text as a list of characters, and access 200 characters starting
|
|
186
|
-
from position 10,000.
|
|
187
|
-
|
|
188
|
-
>>> raw_text = machado.raw('romance/marm05.txt')
|
|
189
|
-
>>> raw_text[10000:10200]
|
|
190
|
-
u', primou no\nEstado, e foi um dos amigos particulares do vice-rei Conde
|
|
191
|
-
da Cunha.\n\nComo este apelido de Cubas lhe\ncheirasse excessivamente a
|
|
192
|
-
tanoaria, alegava meu pai, bisneto de Dami\xe3o, que o\ndito ape'
|
|
193
|
-
|
|
194
|
-
However, this is not a very useful way to work with a text. We generally think
|
|
195
|
-
of a text as a sequence of words and punctuation, not characters:
|
|
196
|
-
|
|
197
|
-
>>> text1 = machado.words('romance/marm05.txt')
|
|
198
|
-
>>> text1
|
|
199
|
-
['Romance', ',', 'Mem\xf3rias', 'P\xf3stumas', 'de', ...]
|
|
200
|
-
>>> len(text1)
|
|
201
|
-
77098
|
|
202
|
-
>>> len(set(text1))
|
|
203
|
-
10848
|
|
204
|
-
|
|
205
|
-
Here's a program that finds the most common ngrams that contain a
|
|
206
|
-
particular target word.
|
|
207
|
-
|
|
208
|
-
>>> from nltk import ngrams, FreqDist
|
|
209
|
-
>>> target_word = 'olhos'
|
|
210
|
-
>>> fd = FreqDist(ng
|
|
211
|
-
... for ng in ngrams(text1, 5)
|
|
212
|
-
... if target_word in ng)
|
|
213
|
-
>>> for hit in fd.samples():
|
|
214
|
-
... print(' '.join(hit))
|
|
215
|
-
...
|
|
216
|
-
, com os olhos no
|
|
217
|
-
com os olhos no ar
|
|
218
|
-
com os olhos no chão
|
|
219
|
-
e todos com os olhos
|
|
220
|
-
me estar com os olhos
|
|
221
|
-
os olhos estúpidos , a
|
|
222
|
-
os olhos na costura ,
|
|
223
|
-
os olhos no ar ,
|
|
224
|
-
, com os olhos espetados
|
|
225
|
-
, com os olhos estúpidos
|
|
226
|
-
, com os olhos fitos
|
|
227
|
-
, com os olhos naquele
|
|
228
|
-
, com os olhos para
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
Accessing the MacMorpho Tagged Corpus
|
|
232
|
-
-------------------------------------
|
|
233
|
-
|
|
234
|
-
NLTK includes the MAC-MORPHO Brazilian Portuguese POS-tagged news text,
|
|
235
|
-
with over a million words of
|
|
236
|
-
journalistic texts extracted from ten sections of
|
|
237
|
-
the daily newspaper *Folha de Sao Paulo*, 1994.
|
|
238
|
-
|
|
239
|
-
We can access this corpus as a sequence of words or tagged words as follows:
|
|
240
|
-
>>> import nltk.corpus
|
|
241
|
-
>>> nltk.corpus.mac_morpho.words()
|
|
242
|
-
['Jersei', 'atinge', 'm\xe9dia', 'de', 'Cr$', '1,4', ...]
|
|
243
|
-
>>> nltk.corpus.mac_morpho.sents() # doctest: +NORMALIZE_WHITESPACE
|
|
244
|
-
[['Jersei', 'atinge', 'm\xe9dia', 'de', 'Cr$', '1,4', 'milh\xe3o',
|
|
245
|
-
'em', 'a', 'venda', 'de', 'a', 'Pinhal', 'em', 'S\xe3o', 'Paulo'],
|
|
246
|
-
['Programe', 'sua', 'viagem', 'a', 'a', 'Exposi\xe7\xe3o', 'Nacional',
|
|
247
|
-
'do', 'Zebu', ',', 'que', 'come\xe7a', 'dia', '25'], ...]
|
|
248
|
-
>>> nltk.corpus.mac_morpho.tagged_words()
|
|
249
|
-
[('Jersei', 'N'), ('atinge', 'V'), ('m\xe9dia', 'N'), ...]
|
|
250
|
-
|
|
251
|
-
We can also access it in sentence chunks.
|
|
252
|
-
|
|
253
|
-
>>> nltk.corpus.mac_morpho.tagged_sents() # doctest: +NORMALIZE_WHITESPACE
|
|
254
|
-
[[('Jersei', 'N'), ('atinge', 'V'), ('m\xe9dia', 'N'), ('de', 'PREP'),
|
|
255
|
-
('Cr$', 'CUR'), ('1,4', 'NUM'), ('milh\xe3o', 'N'), ('em', 'PREP|+'),
|
|
256
|
-
('a', 'ART'), ('venda', 'N'), ('de', 'PREP|+'), ('a', 'ART'),
|
|
257
|
-
('Pinhal', 'NPROP'), ('em', 'PREP'), ('S\xe3o', 'NPROP'),
|
|
258
|
-
('Paulo', 'NPROP')],
|
|
259
|
-
[('Programe', 'V'), ('sua', 'PROADJ'), ('viagem', 'N'), ('a', 'PREP|+'),
|
|
260
|
-
('a', 'ART'), ('Exposi\xe7\xe3o', 'NPROP'), ('Nacional', 'NPROP'),
|
|
261
|
-
('do', 'NPROP'), ('Zebu', 'NPROP'), (',', ','), ('que', 'PRO-KS-REL'),
|
|
262
|
-
('come\xe7a', 'V'), ('dia', 'N'), ('25', 'N|AP')], ...]
|
|
263
|
-
|
|
264
|
-
This data can be used to train taggers (examples below for the Floresta treebank).
|
|
265
|
-
|
|
266
|
-
Accessing the Floresta Portuguese Treebank
|
|
267
|
-
------------------------------------------
|
|
268
|
-
|
|
269
|
-
The NLTK data distribution includes the
|
|
270
|
-
"Floresta Sinta(c)tica Corpus" version 7.4, available from
|
|
271
|
-
``http://www.linguateca.pt/Floresta/``.
|
|
272
|
-
|
|
273
|
-
We can access this corpus as a sequence of words or tagged words as follows:
|
|
274
|
-
|
|
275
|
-
>>> from nltk.corpus import floresta
|
|
276
|
-
>>> floresta.words()
|
|
277
|
-
['Um', 'revivalismo', 'refrescante', 'O', '7_e_Meio', ...]
|
|
278
|
-
>>> floresta.tagged_words()
|
|
279
|
-
[('Um', '>N+art'), ('revivalismo', 'H+n'), ...]
|
|
280
|
-
|
|
281
|
-
The tags consist of some syntactic information, followed by a plus sign,
|
|
282
|
-
followed by a conventional part-of-speech tag. Let's strip off the material before
|
|
283
|
-
the plus sign:
|
|
284
|
-
|
|
285
|
-
>>> def simplify_tag(t):
|
|
286
|
-
... if "+" in t:
|
|
287
|
-
... return t[t.index("+")+1:]
|
|
288
|
-
... else:
|
|
289
|
-
... return t
|
|
290
|
-
>>> twords = floresta.tagged_words()
|
|
291
|
-
>>> twords = [(w.lower(), simplify_tag(t)) for (w,t) in twords]
|
|
292
|
-
>>> twords[:10]
|
|
293
|
-
[('um', 'art'), ('revivalismo', 'n'), ('refrescante', 'adj'), ('o', 'art'), ('7_e_meio', 'prop'),
|
|
294
|
-
('\xe9', 'v-fin'), ('um', 'art'), ('ex-libris', 'n'), ('de', 'prp'), ('a', 'art')]
|
|
295
|
-
|
|
296
|
-
Pretty printing the tagged words:
|
|
297
|
-
|
|
298
|
-
>>> print(' '.join(word + '/' + tag for (word, tag) in twords[:10]))
|
|
299
|
-
um/art revivalismo/n refrescante/adj o/art 7_e_meio/prop é/v-fin um/art ex-libris/n de/prp a/art
|
|
300
|
-
|
|
301
|
-
Count the word tokens and types, and determine the most common word:
|
|
302
|
-
|
|
303
|
-
>>> words = floresta.words()
|
|
304
|
-
>>> len(words)
|
|
305
|
-
211852
|
|
306
|
-
>>> fd = nltk.FreqDist(words)
|
|
307
|
-
>>> len(fd)
|
|
308
|
-
29421
|
|
309
|
-
>>> fd.max()
|
|
310
|
-
'de'
|
|
311
|
-
|
|
312
|
-
List the 20 most frequent tags, in order of decreasing frequency:
|
|
313
|
-
|
|
314
|
-
>>> tags = [simplify_tag(tag) for (word,tag) in floresta.tagged_words()]
|
|
315
|
-
>>> fd = nltk.FreqDist(tags)
|
|
316
|
-
>>> fd.keys()[:20] # doctest: +NORMALIZE_WHITESPACE
|
|
317
|
-
['n', 'prp', 'art', 'v-fin', ',', 'prop', 'adj', 'adv', '.',
|
|
318
|
-
'conj-c', 'v-inf', 'pron-det', 'v-pcp', 'num', 'pron-indp',
|
|
319
|
-
'pron-pers', '\xab', '\xbb', 'conj-s', '}']
|
|
320
|
-
|
|
321
|
-
We can also access the corpus grouped by sentence:
|
|
322
|
-
|
|
323
|
-
>>> floresta.sents() # doctest: +NORMALIZE_WHITESPACE
|
|
324
|
-
[['Um', 'revivalismo', 'refrescante'],
|
|
325
|
-
['O', '7_e_Meio', '\xe9', 'um', 'ex-libris', 'de', 'a', 'noite',
|
|
326
|
-
'algarvia', '.'], ...]
|
|
327
|
-
>>> floresta.tagged_sents() # doctest: +NORMALIZE_WHITESPACE
|
|
328
|
-
[[('Um', '>N+art'), ('revivalismo', 'H+n'), ('refrescante', 'N<+adj')],
|
|
329
|
-
[('O', '>N+art'), ('7_e_Meio', 'H+prop'), ('\xe9', 'P+v-fin'),
|
|
330
|
-
('um', '>N+art'), ('ex-libris', 'H+n'), ('de', 'H+prp'),
|
|
331
|
-
('a', '>N+art'), ('noite', 'H+n'), ('algarvia', 'N<+adj'), ('.', '.')],
|
|
332
|
-
...]
|
|
333
|
-
>>> floresta.parsed_sents() # doctest: +NORMALIZE_WHITESPACE +ELLIPSIS
|
|
334
|
-
[Tree('UTT+np', [Tree('>N+art', ['Um']), Tree('H+n', ['revivalismo']),
|
|
335
|
-
Tree('N<+adj', ['refrescante'])]),
|
|
336
|
-
Tree('STA+fcl',
|
|
337
|
-
[Tree('SUBJ+np', [Tree('>N+art', ['O']),
|
|
338
|
-
Tree('H+prop', ['7_e_Meio'])]),
|
|
339
|
-
Tree('P+v-fin', ['\xe9']),
|
|
340
|
-
Tree('SC+np',
|
|
341
|
-
[Tree('>N+art', ['um']),
|
|
342
|
-
Tree('H+n', ['ex-libris']),
|
|
343
|
-
Tree('N<+pp', [Tree('H+prp', ['de']),
|
|
344
|
-
Tree('P<+np', [Tree('>N+art', ['a']),
|
|
345
|
-
Tree('H+n', ['noite']),
|
|
346
|
-
Tree('N<+adj', ['algarvia'])])])]),
|
|
347
|
-
Tree('.', ['.'])]), ...]
|
|
348
|
-
|
|
349
|
-
To view a parse tree, use the ``draw()`` method, e.g.:
|
|
350
|
-
|
|
351
|
-
>>> psents = floresta.parsed_sents()
|
|
352
|
-
>>> psents[5].draw() # doctest: +SKIP
|
|
353
|
-
|
|
354
|
-
Character Encodings
|
|
355
|
-
-------------------
|
|
356
|
-
|
|
357
|
-
Python understands the common character encoding used for Portuguese, ISO 8859-1 (ISO Latin 1).
|
|
358
|
-
|
|
359
|
-
>>> import os, nltk.test
|
|
360
|
-
>>> testdir = os.path.split(nltk.test.__file__)[0]
|
|
361
|
-
>>> text = open(os.path.join(testdir, 'floresta.txt'), 'rb').read().decode('ISO 8859-1')
|
|
362
|
-
>>> text[:60]
|
|
363
|
-
'O 7 e Meio \xe9 um ex-libris da noite algarvia.\n\xc9 uma das mais '
|
|
364
|
-
>>> print(text[:60])
|
|
365
|
-
O 7 e Meio é um ex-libris da noite algarvia.
|
|
366
|
-
É uma das mais
|
|
367
|
-
|
|
368
|
-
For more information about character encodings and Python, please see section 3.3 of the book.
|
|
369
|
-
|
|
370
|
-
----------------
|
|
371
|
-
Processing Tasks
|
|
372
|
-
----------------
|
|
373
|
-
|
|
374
|
-
|
|
375
|
-
Simple Concordancing
|
|
376
|
-
--------------------
|
|
377
|
-
|
|
378
|
-
Here's a function that takes a word and a specified amount of context (measured
|
|
379
|
-
in characters), and generates a concordance for that word.
|
|
380
|
-
|
|
381
|
-
>>> def concordance(word, context=30):
|
|
382
|
-
... for sent in floresta.sents():
|
|
383
|
-
... if word in sent:
|
|
384
|
-
... pos = sent.index(word)
|
|
385
|
-
... left = ' '.join(sent[:pos])
|
|
386
|
-
... right = ' '.join(sent[pos+1:])
|
|
387
|
-
... print('%*s %s %-*s' %
|
|
388
|
-
... (context, left[-context:], word, context, right[:context]))
|
|
389
|
-
|
|
390
|
-
>>> concordance("dar") # doctest: +SKIP
|
|
391
|
-
anduru , foi o suficiente para dar a volta a o resultado .
|
|
392
|
-
1. O P?BLICO veio dar a a imprensa di?ria portuguesa
|
|
393
|
-
A fartura de pensamento pode dar maus resultados e n?s n?o quer
|
|
394
|
-
Come?a a dar resultados a pol?tica de a Uni
|
|
395
|
-
ial come?ar a incorporar- lo e dar forma a um ' site ' que tem se
|
|
396
|
-
r com Constantino para ele lhe dar tamb?m os pap?is assinados .
|
|
397
|
-
va a brincar , pois n?o lhe ia dar procura??o nenhuma enquanto n?
|
|
398
|
-
?rica como o ant?doto capaz de dar sentido a o seu enorme poder .
|
|
399
|
-
. . .
|
|
400
|
-
>>> concordance("vender") # doctest: +SKIP
|
|
401
|
-
er recebido uma encomenda para vender 4000 blindados a o Iraque .
|
|
402
|
-
m?rico_Amorim caso conseguisse vender o lote de ac??es de o empres?r
|
|
403
|
-
mpre ter jovens simp?ticos a ? vender ? chega ! }
|
|
404
|
-
Disse que o governo vai vender ? desde autom?vel at? particip
|
|
405
|
-
ndiciou ontem duas pessoas por vender carro com ?gio .
|
|
406
|
-
A inten??o de Fleury ? vender as a??es para equilibrar as fi
|
|
407
|
-
|
|
408
|
-
Part-of-Speech Tagging
|
|
409
|
-
----------------------
|
|
410
|
-
|
|
411
|
-
Let's begin by getting the tagged sentence data, and simplifying the tags
|
|
412
|
-
as described earlier.
|
|
413
|
-
|
|
414
|
-
>>> from nltk.corpus import floresta
|
|
415
|
-
>>> tsents = floresta.tagged_sents()
|
|
416
|
-
>>> tsents = [[(w.lower(),simplify_tag(t)) for (w,t) in sent] for sent in tsents if sent]
|
|
417
|
-
>>> train = tsents[100:]
|
|
418
|
-
>>> test = tsents[:100]
|
|
419
|
-
|
|
420
|
-
We already know that ``n`` is the most common tag, so we can set up a
|
|
421
|
-
default tagger that tags every word as a noun, and see how well it does:
|
|
422
|
-
|
|
423
|
-
>>> tagger0 = nltk.DefaultTagger('n')
|
|
424
|
-
>>> nltk.tag.accuracy(tagger0, test)
|
|
425
|
-
0.17697228144989338
|
|
426
|
-
|
|
427
|
-
Evidently, about one in every six words is a noun. Let's improve on this by
|
|
428
|
-
training a unigram tagger:
|
|
429
|
-
|
|
430
|
-
>>> tagger1 = nltk.UnigramTagger(train, backoff=tagger0)
|
|
431
|
-
>>> nltk.tag.accuracy(tagger1, test)
|
|
432
|
-
0.87029140014214645
|
|
433
|
-
|
|
434
|
-
Next a bigram tagger:
|
|
435
|
-
|
|
436
|
-
>>> tagger2 = nltk.BigramTagger(train, backoff=tagger1)
|
|
437
|
-
>>> nltk.tag.accuracy(tagger2, test)
|
|
438
|
-
0.89019189765458417
|
|
439
|
-
|
|
440
|
-
|
|
441
|
-
Sentence Segmentation
|
|
442
|
-
---------------------
|
|
443
|
-
|
|
444
|
-
Punkt is a language-neutral sentence segmentation tool. We
|
|
445
|
-
|
|
446
|
-
>>> sent_tokenizer=nltk.data.load('tokenizers/punkt/portuguese.pickle')
|
|
447
|
-
>>> raw_text = machado.raw('romance/marm05.txt')
|
|
448
|
-
>>> sentences = sent_tokenizer.tokenize(raw_text)
|
|
449
|
-
>>> for sent in sentences[1000:1005]:
|
|
450
|
-
... print("<<", sent, ">>")
|
|
451
|
-
...
|
|
452
|
-
<< Em verdade, parecia ainda mais mulher do que era;
|
|
453
|
-
seria criança nos seus folgares de moça; mas assim quieta, impassível, tinha a
|
|
454
|
-
compostura da mulher casada. >>
|
|
455
|
-
<< Talvez essa circunstância lhe diminuía um pouco da
|
|
456
|
-
graça virginal. >>
|
|
457
|
-
<< Depressa nos familiarizamos; a mãe fazia-lhe grandes elogios, eu
|
|
458
|
-
escutava-os de boa sombra, e ela sorria com os olhos fúlgidos, como se lá dentro
|
|
459
|
-
do cérebro lhe estivesse a voar uma borboletinha de asas de ouro e olhos de
|
|
460
|
-
diamante... >>
|
|
461
|
-
<< Digo lá dentro, porque cá fora o
|
|
462
|
-
que esvoaçou foi uma borboleta preta, que subitamente penetrou na varanda, e
|
|
463
|
-
começou a bater as asas em derredor de D. Eusébia. >>
|
|
464
|
-
<< D. Eusébia deu um grito,
|
|
465
|
-
levantou-se, praguejou umas palavras soltas: - T'esconjuro!... >>
|
|
466
|
-
|
|
467
|
-
The sentence tokenizer can be trained and evaluated on other text.
|
|
468
|
-
The source text (from the Floresta Portuguese Treebank) contains one sentence per line.
|
|
469
|
-
We read the text, split it into its lines, and then join these lines together using
|
|
470
|
-
spaces. Now the information about sentence breaks has been discarded. We split this
|
|
471
|
-
material into training and testing data:
|
|
472
|
-
|
|
473
|
-
>>> import os, nltk.test
|
|
474
|
-
>>> testdir = os.path.split(nltk.test.__file__)[0]
|
|
475
|
-
>>> text = open(os.path.join(testdir, 'floresta.txt'), 'rb').read().decode('ISO-8859-1')
|
|
476
|
-
>>> lines = text.split('\n')
|
|
477
|
-
>>> train = ' '.join(lines[10:])
|
|
478
|
-
>>> test = ' '.join(lines[:10])
|
|
479
|
-
|
|
480
|
-
Now we train the sentence segmenter (or sentence tokenizer) and use it on our test sentences:
|
|
481
|
-
|
|
482
|
-
>>> stok = nltk.PunktSentenceTokenizer(train)
|
|
483
|
-
>>> print(stok.tokenize(test))
|
|
484
|
-
['O 7 e Meio \xe9 um ex-libris da noite algarvia.',
|
|
485
|
-
'\xc9 uma das mais antigas discotecas do Algarve, situada em Albufeira,
|
|
486
|
-
que continua a manter os tra\xe7os decorativos e as clientelas de sempre.',
|
|
487
|
-
'\xc9 um pouco a vers\xe3o de uma esp\xe9cie de \xaboutro lado\xbb da noite,
|
|
488
|
-
a meio caminho entre os devaneios de uma fauna perif\xe9rica, seja de Lisboa,
|
|
489
|
-
Londres, Dublin ou Faro e Portim\xe3o, e a postura circunspecta dos fi\xe9is da casa,
|
|
490
|
-
que dela esperam a m\xfasica \xabgeracionista\xbb dos 60 ou dos 70.',
|
|
491
|
-
'N\xe3o deixa de ser, nos tempos que correm, um certo \xabvery typical\xbb algarvio,
|
|
492
|
-
cabe\xe7a de cartaz para os que querem fugir a algumas movimenta\xe7\xf5es nocturnas
|
|
493
|
-
j\xe1 a caminho da ritualiza\xe7\xe3o de massas, do g\xe9nero \xabvamos todos ao
|
|
494
|
-
Calypso e encontramo-nos na Locomia\xbb.',
|
|
495
|
-
'E assim, aos 2,5 milh\xf5es que o Minist\xe9rio do Planeamento e Administra\xe7\xe3o
|
|
496
|
-
do Territ\xf3rio j\xe1 gasta no pagamento do pessoal afecto a estes organismos,
|
|
497
|
-
v\xeam juntar-se os montantes das obras propriamente ditas, que os munic\xedpios,
|
|
498
|
-
j\xe1 com projectos na m\xe3o, v\xeam reivindicar junto do Executivo, como salienta
|
|
499
|
-
aquele membro do Governo.',
|
|
500
|
-
'E o dinheiro \xabn\xe3o falta s\xf3 \xe0s c\xe2maras\xbb, lembra o secret\xe1rio de Estado,
|
|
501
|
-
que considera que a solu\xe7\xe3o para as autarquias \xe9 \xabespecializarem-se em
|
|
502
|
-
fundos comunit\xe1rios\xbb.',
|
|
503
|
-
'Mas como, se muitas n\xe3o disp\xf5em, nos seus quadros, dos t\xe9cnicos necess\xe1rios?',
|
|
504
|
-
'\xabEncomendem-nos a projectistas de fora\xbb porque, se as obras vierem a ser financiadas,
|
|
505
|
-
eles at\xe9 saem de gra\xe7a, j\xe1 que, nesse caso, \xabos fundos comunit\xe1rios pagam
|
|
506
|
-
os projectos, o mesmo n\xe3o acontecendo quando eles s\xe3o feitos pelos GAT\xbb,
|
|
507
|
-
dado serem organismos do Estado.',
|
|
508
|
-
'Essa poder\xe1 vir a ser uma hip\xf3tese, at\xe9 porque, no terreno, a capacidade dos GAT
|
|
509
|
-
est\xe1 cada vez mais enfraquecida.',
|
|
510
|
-
'Alguns at\xe9 j\xe1 desapareceram, como o de Castro Verde, e outros t\xeam vindo a perder quadros.']
|
|
511
|
-
|
|
512
|
-
NLTK's data collection includes a trained model for Portuguese sentence
|
|
513
|
-
segmentation, which can be loaded as follows. It is faster to load a trained model than
|
|
514
|
-
to retrain it.
|
|
515
|
-
|
|
516
|
-
>>> stok = nltk.data.load('tokenizers/punkt/portuguese.pickle')
|
|
517
|
-
|
|
518
|
-
Stemming
|
|
519
|
-
--------
|
|
520
|
-
|
|
521
|
-
NLTK includes the RSLP Portuguese stemmer. Here we use it to stem some Portuguese text:
|
|
522
|
-
|
|
523
|
-
>>> stemmer = nltk.stem.RSLPStemmer()
|
|
524
|
-
>>> stemmer.stem("copiar")
|
|
525
|
-
'copi'
|
|
526
|
-
>>> stemmer.stem("paisagem")
|
|
527
|
-
'pais'
|
|
528
|
-
|
|
529
|
-
|
|
530
|
-
Stopwords
|
|
531
|
-
---------
|
|
532
|
-
|
|
533
|
-
NLTK includes Portuguese stopwords:
|
|
534
|
-
|
|
535
|
-
>>> stopwords = nltk.corpus.stopwords.words('portuguese')
|
|
536
|
-
>>> stopwords[:10]
|
|
537
|
-
['a', 'ao', 'aos', 'aquela', 'aquelas', 'aquele', 'aqueles', 'aquilo', 'as', 'at\xe9']
|
|
538
|
-
|
|
539
|
-
Now we can use these to filter text. Let's find the most frequent words (other than stopwords)
|
|
540
|
-
and print them in descending order of frequency:
|
|
541
|
-
|
|
542
|
-
>>> fd = nltk.FreqDist(w.lower() for w in floresta.words() if w not in stopwords)
|
|
543
|
-
>>> for word in list(fd.keys())[:20]:
|
|
544
|
-
... print(word, fd[word])
|
|
545
|
-
, 13444
|
|
546
|
-
. 7725
|
|
547
|
-
« 2369
|
|
548
|
-
» 2310
|
|
549
|
-
é 1305
|
|
550
|
-
o 1086
|
|
551
|
-
} 1047
|
|
552
|
-
{ 1044
|
|
553
|
-
a 897
|
|
554
|
-
; 633
|
|
555
|
-
em 516
|
|
556
|
-
ser 466
|
|
557
|
-
sobre 349
|
|
558
|
-
os 313
|
|
559
|
-
anos 301
|
|
560
|
-
ontem 292
|
|
561
|
-
ainda 279
|
|
562
|
-
segundo 256
|
|
563
|
-
ter 249
|
|
564
|
-
dois 231
|
|
565
|
-
|
|
1
|
+
.. Copyright (C) 2001-2018 NLTK Project
|
|
2
|
+
.. For license information, see LICENSE.TXT
|
|
3
|
+
|
|
4
|
+
==================================
|
|
5
|
+
Examples for Portuguese Processing
|
|
6
|
+
==================================
|
|
7
|
+
|
|
8
|
+
This HOWTO contains a variety of examples relating to the Portuguese language.
|
|
9
|
+
It is intended to be read in conjunction with the NLTK book
|
|
10
|
+
(``http://nltk.org/book``). For instructions on running the Python
|
|
11
|
+
interpreter, please see the section *Getting Started with Python*, in Chapter 1.
|
|
12
|
+
|
|
13
|
+
--------------------------------------------
|
|
14
|
+
Python Programming, with Portuguese Examples
|
|
15
|
+
--------------------------------------------
|
|
16
|
+
|
|
17
|
+
Chapter 1 of the NLTK book contains many elementary programming examples, all
|
|
18
|
+
with English texts. In this section, we'll see some corresponding examples
|
|
19
|
+
using Portuguese. Please refer to the chapter for full discussion. *Vamos!*
|
|
20
|
+
|
|
21
|
+
>>> from nltk.examples.pt import *
|
|
22
|
+
*** Introductory Examples for the NLTK Book ***
|
|
23
|
+
Loading ptext1, ... and psent1, ...
|
|
24
|
+
Type the name of the text or sentence to view it.
|
|
25
|
+
Type: 'texts()' or 'sents()' to list the materials.
|
|
26
|
+
ptext1: Memórias Póstumas de Brás Cubas (1881)
|
|
27
|
+
ptext2: Dom Casmurro (1899)
|
|
28
|
+
ptext3: Gênesis
|
|
29
|
+
ptext4: Folha de Sao Paulo (1994)
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
Any time we want to find out about these texts, we just have
|
|
33
|
+
to enter their names at the Python prompt:
|
|
34
|
+
|
|
35
|
+
>>> ptext2
|
|
36
|
+
<Text: Dom Casmurro (1899)>
|
|
37
|
+
|
|
38
|
+
Searching Text
|
|
39
|
+
--------------
|
|
40
|
+
|
|
41
|
+
A concordance permits us to see words in context.
|
|
42
|
+
|
|
43
|
+
>>> ptext1.concordance('olhos')
|
|
44
|
+
Building index...
|
|
45
|
+
Displaying 25 of 138 matches:
|
|
46
|
+
De pé , à cabeceira da cama , com os olhos estúpidos , a boca entreaberta , a t
|
|
47
|
+
orelhas . Pela minha parte fechei os olhos e deixei - me ir à ventura . Já agor
|
|
48
|
+
xões de cérebro enfermo . Como ia de olhos fechados , não via o caminho ; lembr
|
|
49
|
+
gelos eternos . Com efeito , abri os olhos e vi que o meu animal galopava numa
|
|
50
|
+
me apareceu então , fitando - me uns olhos rutilantes como o sol . Tudo nessa f
|
|
51
|
+
mim mesmo . Então , encarei - a com olhos súplices , e pedi mais alguns anos .
|
|
52
|
+
...
|
|
53
|
+
|
|
54
|
+
For a given word, we can find words with a similar text distribution:
|
|
55
|
+
|
|
56
|
+
>>> ptext1.similar('chegar')
|
|
57
|
+
Building word-context index...
|
|
58
|
+
acabada acudir aludir avistar bramanismo casamento cheguei com contar
|
|
59
|
+
contrário corpo dali deixei desferirem dizer fazer filhos já leitor lhe
|
|
60
|
+
>>> ptext3.similar('chegar')
|
|
61
|
+
Building word-context index...
|
|
62
|
+
achar alumiar arrombar destruir governar guardar ir lavrar passar que
|
|
63
|
+
toda tomar ver vir
|
|
64
|
+
|
|
65
|
+
We can search for the statistically significant collocations in a text:
|
|
66
|
+
|
|
67
|
+
>>> ptext1.collocations()
|
|
68
|
+
Building collocations list
|
|
69
|
+
Quincas Borba; Lobo Neves; alguma coisa; Brás Cubas; meu pai; dia
|
|
70
|
+
seguinte; não sei; Meu pai; alguns instantes; outra vez; outra coisa;
|
|
71
|
+
por exemplo; mim mesmo; coisa nenhuma; mesma coisa; não era; dias
|
|
72
|
+
depois; Passeio Público; olhar para; das coisas
|
|
73
|
+
|
|
74
|
+
We can search for words in context, with the help of *regular expressions*, e.g.:
|
|
75
|
+
|
|
76
|
+
>>> ptext1.findall("<olhos> (<.*>)")
|
|
77
|
+
estúpidos; e; fechados; rutilantes; súplices; a; do; babavam;
|
|
78
|
+
na; moles; se; da; umas; espraiavam; chamejantes; espetados;
|
|
79
|
+
...
|
|
80
|
+
|
|
81
|
+
We can automatically generate random text based on a given text, e.g.:
|
|
82
|
+
|
|
83
|
+
>>> ptext3.generate() # doctest: +SKIP
|
|
84
|
+
No princípio , criou Deus os abençoou , dizendo : Onde { estão } e até
|
|
85
|
+
à ave dos céus , { que } será . Disse mais Abrão : Dá - me a mulher
|
|
86
|
+
que tomaste ; porque daquele poço Eseque , { tinha .} E disse : Não
|
|
87
|
+
poderemos descer ; mas , do campo ainda não estava na casa do teu
|
|
88
|
+
pescoço . E viveu Serugue , depois Simeão e Levi { são } estes ? E o
|
|
89
|
+
varão , porque habitava na terra de Node , da mão de Esaú : Jeús ,
|
|
90
|
+
Jalão e Corá
|
|
91
|
+
|
|
92
|
+
Texts as List of Words
|
|
93
|
+
----------------------
|
|
94
|
+
|
|
95
|
+
A few sentences have been defined for you.
|
|
96
|
+
|
|
97
|
+
>>> psent1
|
|
98
|
+
['o', 'amor', 'da', 'gl\xf3ria', 'era', 'a', 'coisa', 'mais',
|
|
99
|
+
'verdadeiramente', 'humana', 'que', 'h\xe1', 'no', 'homem', ',',
|
|
100
|
+
'e', ',', 'conseq\xfcentemente', ',', 'a', 'sua', 'mais',
|
|
101
|
+
'genu\xedna', 'fei\xe7\xe3o', '.']
|
|
102
|
+
>>>
|
|
103
|
+
|
|
104
|
+
Notice that the sentence has been *tokenized*. Each token is
|
|
105
|
+
represented as a string, represented using quotes, e.g. ``'coisa'``.
|
|
106
|
+
Some strings contain special characters, e.g. ``\xf3``,
|
|
107
|
+
the internal representation for ó.
|
|
108
|
+
The tokens are combined in the form of a *list*. How long is this list?
|
|
109
|
+
|
|
110
|
+
>>> len(psent1)
|
|
111
|
+
25
|
|
112
|
+
>>>
|
|
113
|
+
|
|
114
|
+
What is the vocabulary of this sentence?
|
|
115
|
+
|
|
116
|
+
>>> sorted(set(psent1))
|
|
117
|
+
[',', '.', 'a', 'amor', 'coisa', 'conseqüentemente', 'da', 'e', 'era',
|
|
118
|
+
'feição', 'genuína', 'glória', 'homem', 'humana', 'há', 'mais', 'no',
|
|
119
|
+
'o', 'que', 'sua', 'verdadeiramente']
|
|
120
|
+
>>>
|
|
121
|
+
|
|
122
|
+
Let's iterate over each item in ``psent2``, and print information for each:
|
|
123
|
+
|
|
124
|
+
>>> for w in psent2:
|
|
125
|
+
... print(w, len(w), w[-1])
|
|
126
|
+
...
|
|
127
|
+
Não 3 o
|
|
128
|
+
consultes 9 s
|
|
129
|
+
dicionários 11 s
|
|
130
|
+
. 1 .
|
|
131
|
+
|
|
132
|
+
Observe how we make a human-readable version of a string, using ``decode()``.
|
|
133
|
+
Also notice that we accessed the last character of a string ``w`` using ``w[-1]``.
|
|
134
|
+
|
|
135
|
+
We just saw a ``for`` loop above. Another useful control structure is a
|
|
136
|
+
*list comprehension*.
|
|
137
|
+
|
|
138
|
+
>>> [w.upper() for w in psent2]
|
|
139
|
+
['N\xc3O', 'CONSULTES', 'DICION\xc1RIOS', '.']
|
|
140
|
+
>>> [w for w in psent1 if w.endswith('a')]
|
|
141
|
+
['da', 'gl\xf3ria', 'era', 'a', 'coisa', 'humana', 'a', 'sua', 'genu\xedna']
|
|
142
|
+
>>> [w for w in ptext4 if len(w) > 15]
|
|
143
|
+
[u'norte-irlandeses', u'pan-nacionalismo', u'predominatemente', u'primeiro-ministro',
|
|
144
|
+
u'primeiro-ministro', u'irlandesa-americana', u'responsabilidades', u'significativamente']
|
|
145
|
+
|
|
146
|
+
We can examine the relative frequency of words in a text, using ``FreqDist``:
|
|
147
|
+
|
|
148
|
+
>>> fd1 = FreqDist(ptext1)
|
|
149
|
+
>>> fd1
|
|
150
|
+
<FreqDist with 10848 samples and 77098 outcomes>
|
|
151
|
+
>>> fd1['olhos']
|
|
152
|
+
137
|
|
153
|
+
>>> fd1.max()
|
|
154
|
+
u','
|
|
155
|
+
>>> fd1.samples()[:100]
|
|
156
|
+
[u',', u'.', u'a', u'que', u'de', u'e', u'-', u'o', u';', u'me', u'um', u'n\xe3o',
|
|
157
|
+
u'\x97', u'se', u'do', u'da', u'uma', u'com', u'os', u'\xe9', u'era', u'as', u'eu',
|
|
158
|
+
u'lhe', u'ao', u'em', u'para', u'mas', u'...', u'!', u'\xe0', u'na', u'mais', u'?',
|
|
159
|
+
u'no', u'como', u'por', u'N\xe3o', u'dos', u'ou', u'ele', u':', u'Virg\xedlia',
|
|
160
|
+
u'meu', u'disse', u'minha', u'das', u'O', u'/', u'A', u'CAP\xcdTULO', u'muito',
|
|
161
|
+
u'depois', u'coisa', u'foi', u'sem', u'olhos', u'ela', u'nos', u'tinha', u'nem',
|
|
162
|
+
u'E', u'outro', u'vida', u'nada', u'tempo', u'menos', u'outra', u'casa', u'homem',
|
|
163
|
+
u'porque', u'quando', u'mim', u'mesmo', u'ser', u'pouco', u'estava', u'dia',
|
|
164
|
+
u't\xe3o', u'tudo', u'Mas', u'at\xe9', u'D', u'ainda', u's\xf3', u'alguma',
|
|
165
|
+
u'la', u'vez', u'anos', u'h\xe1', u'Era', u'pai', u'esse', u'lo', u'dizer', u'assim',
|
|
166
|
+
u'ent\xe3o', u'dizia', u'aos', u'Borba']
|
|
167
|
+
|
|
168
|
+
---------------
|
|
169
|
+
Reading Corpora
|
|
170
|
+
---------------
|
|
171
|
+
|
|
172
|
+
Accessing the Machado Text Corpus
|
|
173
|
+
---------------------------------
|
|
174
|
+
|
|
175
|
+
NLTK includes the complete works of Machado de Assis.
|
|
176
|
+
|
|
177
|
+
>>> from nltk.corpus import machado
|
|
178
|
+
>>> machado.fileids()
|
|
179
|
+
['contos/macn001.txt', 'contos/macn002.txt', 'contos/macn003.txt', ...]
|
|
180
|
+
|
|
181
|
+
Each file corresponds to one of the works of Machado de Assis. To see a complete
|
|
182
|
+
list of works, you can look at the corpus README file: ``print machado.readme()``.
|
|
183
|
+
Let's access the text of the *Posthumous Memories of Brás Cubas*.
|
|
184
|
+
|
|
185
|
+
We can access the text as a list of characters, and access 200 characters starting
|
|
186
|
+
from position 10,000.
|
|
187
|
+
|
|
188
|
+
>>> raw_text = machado.raw('romance/marm05.txt')
|
|
189
|
+
>>> raw_text[10000:10200]
|
|
190
|
+
u', primou no\nEstado, e foi um dos amigos particulares do vice-rei Conde
|
|
191
|
+
da Cunha.\n\nComo este apelido de Cubas lhe\ncheirasse excessivamente a
|
|
192
|
+
tanoaria, alegava meu pai, bisneto de Dami\xe3o, que o\ndito ape'
|
|
193
|
+
|
|
194
|
+
However, this is not a very useful way to work with a text. We generally think
|
|
195
|
+
of a text as a sequence of words and punctuation, not characters:
|
|
196
|
+
|
|
197
|
+
>>> text1 = machado.words('romance/marm05.txt')
|
|
198
|
+
>>> text1
|
|
199
|
+
['Romance', ',', 'Mem\xf3rias', 'P\xf3stumas', 'de', ...]
|
|
200
|
+
>>> len(text1)
|
|
201
|
+
77098
|
|
202
|
+
>>> len(set(text1))
|
|
203
|
+
10848
|
|
204
|
+
|
|
205
|
+
Here's a program that finds the most common ngrams that contain a
|
|
206
|
+
particular target word.
|
|
207
|
+
|
|
208
|
+
>>> from nltk import ngrams, FreqDist
|
|
209
|
+
>>> target_word = 'olhos'
|
|
210
|
+
>>> fd = FreqDist(ng
|
|
211
|
+
... for ng in ngrams(text1, 5)
|
|
212
|
+
... if target_word in ng)
|
|
213
|
+
>>> for hit in fd.samples():
|
|
214
|
+
... print(' '.join(hit))
|
|
215
|
+
...
|
|
216
|
+
, com os olhos no
|
|
217
|
+
com os olhos no ar
|
|
218
|
+
com os olhos no chão
|
|
219
|
+
e todos com os olhos
|
|
220
|
+
me estar com os olhos
|
|
221
|
+
os olhos estúpidos , a
|
|
222
|
+
os olhos na costura ,
|
|
223
|
+
os olhos no ar ,
|
|
224
|
+
, com os olhos espetados
|
|
225
|
+
, com os olhos estúpidos
|
|
226
|
+
, com os olhos fitos
|
|
227
|
+
, com os olhos naquele
|
|
228
|
+
, com os olhos para
|
|
229
|
+
|
|
230
|
+
|
|
231
|
+
Accessing the MacMorpho Tagged Corpus
|
|
232
|
+
-------------------------------------
|
|
233
|
+
|
|
234
|
+
NLTK includes the MAC-MORPHO Brazilian Portuguese POS-tagged news text,
|
|
235
|
+
with over a million words of
|
|
236
|
+
journalistic texts extracted from ten sections of
|
|
237
|
+
the daily newspaper *Folha de Sao Paulo*, 1994.
|
|
238
|
+
|
|
239
|
+
We can access this corpus as a sequence of words or tagged words as follows:
|
|
240
|
+
>>> import nltk.corpus
|
|
241
|
+
>>> nltk.corpus.mac_morpho.words()
|
|
242
|
+
['Jersei', 'atinge', 'm\xe9dia', 'de', 'Cr$', '1,4', ...]
|
|
243
|
+
>>> nltk.corpus.mac_morpho.sents() # doctest: +NORMALIZE_WHITESPACE
|
|
244
|
+
[['Jersei', 'atinge', 'm\xe9dia', 'de', 'Cr$', '1,4', 'milh\xe3o',
|
|
245
|
+
'em', 'a', 'venda', 'de', 'a', 'Pinhal', 'em', 'S\xe3o', 'Paulo'],
|
|
246
|
+
['Programe', 'sua', 'viagem', 'a', 'a', 'Exposi\xe7\xe3o', 'Nacional',
|
|
247
|
+
'do', 'Zebu', ',', 'que', 'come\xe7a', 'dia', '25'], ...]
|
|
248
|
+
>>> nltk.corpus.mac_morpho.tagged_words()
|
|
249
|
+
[('Jersei', 'N'), ('atinge', 'V'), ('m\xe9dia', 'N'), ...]
|
|
250
|
+
|
|
251
|
+
We can also access it in sentence chunks.
|
|
252
|
+
|
|
253
|
+
>>> nltk.corpus.mac_morpho.tagged_sents() # doctest: +NORMALIZE_WHITESPACE
|
|
254
|
+
[[('Jersei', 'N'), ('atinge', 'V'), ('m\xe9dia', 'N'), ('de', 'PREP'),
|
|
255
|
+
('Cr$', 'CUR'), ('1,4', 'NUM'), ('milh\xe3o', 'N'), ('em', 'PREP|+'),
|
|
256
|
+
('a', 'ART'), ('venda', 'N'), ('de', 'PREP|+'), ('a', 'ART'),
|
|
257
|
+
('Pinhal', 'NPROP'), ('em', 'PREP'), ('S\xe3o', 'NPROP'),
|
|
258
|
+
('Paulo', 'NPROP')],
|
|
259
|
+
[('Programe', 'V'), ('sua', 'PROADJ'), ('viagem', 'N'), ('a', 'PREP|+'),
|
|
260
|
+
('a', 'ART'), ('Exposi\xe7\xe3o', 'NPROP'), ('Nacional', 'NPROP'),
|
|
261
|
+
('do', 'NPROP'), ('Zebu', 'NPROP'), (',', ','), ('que', 'PRO-KS-REL'),
|
|
262
|
+
('come\xe7a', 'V'), ('dia', 'N'), ('25', 'N|AP')], ...]
|
|
263
|
+
|
|
264
|
+
This data can be used to train taggers (examples below for the Floresta treebank).
|
|
265
|
+
|
|
266
|
+
Accessing the Floresta Portuguese Treebank
|
|
267
|
+
------------------------------------------
|
|
268
|
+
|
|
269
|
+
The NLTK data distribution includes the
|
|
270
|
+
"Floresta Sinta(c)tica Corpus" version 7.4, available from
|
|
271
|
+
``http://www.linguateca.pt/Floresta/``.
|
|
272
|
+
|
|
273
|
+
We can access this corpus as a sequence of words or tagged words as follows:
|
|
274
|
+
|
|
275
|
+
>>> from nltk.corpus import floresta
|
|
276
|
+
>>> floresta.words()
|
|
277
|
+
['Um', 'revivalismo', 'refrescante', 'O', '7_e_Meio', ...]
|
|
278
|
+
>>> floresta.tagged_words()
|
|
279
|
+
[('Um', '>N+art'), ('revivalismo', 'H+n'), ...]
|
|
280
|
+
|
|
281
|
+
The tags consist of some syntactic information, followed by a plus sign,
|
|
282
|
+
followed by a conventional part-of-speech tag. Let's strip off the material before
|
|
283
|
+
the plus sign:
|
|
284
|
+
|
|
285
|
+
>>> def simplify_tag(t):
|
|
286
|
+
... if "+" in t:
|
|
287
|
+
... return t[t.index("+")+1:]
|
|
288
|
+
... else:
|
|
289
|
+
... return t
|
|
290
|
+
>>> twords = floresta.tagged_words()
|
|
291
|
+
>>> twords = [(w.lower(), simplify_tag(t)) for (w,t) in twords]
|
|
292
|
+
>>> twords[:10]
|
|
293
|
+
[('um', 'art'), ('revivalismo', 'n'), ('refrescante', 'adj'), ('o', 'art'), ('7_e_meio', 'prop'),
|
|
294
|
+
('\xe9', 'v-fin'), ('um', 'art'), ('ex-libris', 'n'), ('de', 'prp'), ('a', 'art')]
|
|
295
|
+
|
|
296
|
+
Pretty printing the tagged words:
|
|
297
|
+
|
|
298
|
+
>>> print(' '.join(word + '/' + tag for (word, tag) in twords[:10]))
|
|
299
|
+
um/art revivalismo/n refrescante/adj o/art 7_e_meio/prop é/v-fin um/art ex-libris/n de/prp a/art
|
|
300
|
+
|
|
301
|
+
Count the word tokens and types, and determine the most common word:
|
|
302
|
+
|
|
303
|
+
>>> words = floresta.words()
|
|
304
|
+
>>> len(words)
|
|
305
|
+
211852
|
|
306
|
+
>>> fd = nltk.FreqDist(words)
|
|
307
|
+
>>> len(fd)
|
|
308
|
+
29421
|
|
309
|
+
>>> fd.max()
|
|
310
|
+
'de'
|
|
311
|
+
|
|
312
|
+
List the 20 most frequent tags, in order of decreasing frequency:
|
|
313
|
+
|
|
314
|
+
>>> tags = [simplify_tag(tag) for (word,tag) in floresta.tagged_words()]
|
|
315
|
+
>>> fd = nltk.FreqDist(tags)
|
|
316
|
+
>>> fd.keys()[:20] # doctest: +NORMALIZE_WHITESPACE
|
|
317
|
+
['n', 'prp', 'art', 'v-fin', ',', 'prop', 'adj', 'adv', '.',
|
|
318
|
+
'conj-c', 'v-inf', 'pron-det', 'v-pcp', 'num', 'pron-indp',
|
|
319
|
+
'pron-pers', '\xab', '\xbb', 'conj-s', '}']
|
|
320
|
+
|
|
321
|
+
We can also access the corpus grouped by sentence:
|
|
322
|
+
|
|
323
|
+
>>> floresta.sents() # doctest: +NORMALIZE_WHITESPACE
|
|
324
|
+
[['Um', 'revivalismo', 'refrescante'],
|
|
325
|
+
['O', '7_e_Meio', '\xe9', 'um', 'ex-libris', 'de', 'a', 'noite',
|
|
326
|
+
'algarvia', '.'], ...]
|
|
327
|
+
>>> floresta.tagged_sents() # doctest: +NORMALIZE_WHITESPACE
|
|
328
|
+
[[('Um', '>N+art'), ('revivalismo', 'H+n'), ('refrescante', 'N<+adj')],
|
|
329
|
+
[('O', '>N+art'), ('7_e_Meio', 'H+prop'), ('\xe9', 'P+v-fin'),
|
|
330
|
+
('um', '>N+art'), ('ex-libris', 'H+n'), ('de', 'H+prp'),
|
|
331
|
+
('a', '>N+art'), ('noite', 'H+n'), ('algarvia', 'N<+adj'), ('.', '.')],
|
|
332
|
+
...]
|
|
333
|
+
>>> floresta.parsed_sents() # doctest: +NORMALIZE_WHITESPACE +ELLIPSIS
|
|
334
|
+
[Tree('UTT+np', [Tree('>N+art', ['Um']), Tree('H+n', ['revivalismo']),
|
|
335
|
+
Tree('N<+adj', ['refrescante'])]),
|
|
336
|
+
Tree('STA+fcl',
|
|
337
|
+
[Tree('SUBJ+np', [Tree('>N+art', ['O']),
|
|
338
|
+
Tree('H+prop', ['7_e_Meio'])]),
|
|
339
|
+
Tree('P+v-fin', ['\xe9']),
|
|
340
|
+
Tree('SC+np',
|
|
341
|
+
[Tree('>N+art', ['um']),
|
|
342
|
+
Tree('H+n', ['ex-libris']),
|
|
343
|
+
Tree('N<+pp', [Tree('H+prp', ['de']),
|
|
344
|
+
Tree('P<+np', [Tree('>N+art', ['a']),
|
|
345
|
+
Tree('H+n', ['noite']),
|
|
346
|
+
Tree('N<+adj', ['algarvia'])])])]),
|
|
347
|
+
Tree('.', ['.'])]), ...]
|
|
348
|
+
|
|
349
|
+
To view a parse tree, use the ``draw()`` method, e.g.:
|
|
350
|
+
|
|
351
|
+
>>> psents = floresta.parsed_sents()
|
|
352
|
+
>>> psents[5].draw() # doctest: +SKIP
|
|
353
|
+
|
|
354
|
+
Character Encodings
|
|
355
|
+
-------------------
|
|
356
|
+
|
|
357
|
+
Python understands the common character encoding used for Portuguese, ISO 8859-1 (ISO Latin 1).
|
|
358
|
+
|
|
359
|
+
>>> import os, nltk.test
|
|
360
|
+
>>> testdir = os.path.split(nltk.test.__file__)[0]
|
|
361
|
+
>>> text = open(os.path.join(testdir, 'floresta.txt'), 'rb').read().decode('ISO 8859-1')
|
|
362
|
+
>>> text[:60]
|
|
363
|
+
'O 7 e Meio \xe9 um ex-libris da noite algarvia.\n\xc9 uma das mais '
|
|
364
|
+
>>> print(text[:60])
|
|
365
|
+
O 7 e Meio é um ex-libris da noite algarvia.
|
|
366
|
+
É uma das mais
|
|
367
|
+
|
|
368
|
+
For more information about character encodings and Python, please see section 3.3 of the book.
|
|
369
|
+
|
|
370
|
+
----------------
|
|
371
|
+
Processing Tasks
|
|
372
|
+
----------------
|
|
373
|
+
|
|
374
|
+
|
|
375
|
+
Simple Concordancing
|
|
376
|
+
--------------------
|
|
377
|
+
|
|
378
|
+
Here's a function that takes a word and a specified amount of context (measured
|
|
379
|
+
in characters), and generates a concordance for that word.
|
|
380
|
+
|
|
381
|
+
>>> def concordance(word, context=30):
|
|
382
|
+
... for sent in floresta.sents():
|
|
383
|
+
... if word in sent:
|
|
384
|
+
... pos = sent.index(word)
|
|
385
|
+
... left = ' '.join(sent[:pos])
|
|
386
|
+
... right = ' '.join(sent[pos+1:])
|
|
387
|
+
... print('%*s %s %-*s' %
|
|
388
|
+
... (context, left[-context:], word, context, right[:context]))
|
|
389
|
+
|
|
390
|
+
>>> concordance("dar") # doctest: +SKIP
|
|
391
|
+
anduru , foi o suficiente para dar a volta a o resultado .
|
|
392
|
+
1. O P?BLICO veio dar a a imprensa di?ria portuguesa
|
|
393
|
+
A fartura de pensamento pode dar maus resultados e n?s n?o quer
|
|
394
|
+
Come?a a dar resultados a pol?tica de a Uni
|
|
395
|
+
ial come?ar a incorporar- lo e dar forma a um ' site ' que tem se
|
|
396
|
+
r com Constantino para ele lhe dar tamb?m os pap?is assinados .
|
|
397
|
+
va a brincar , pois n?o lhe ia dar procura??o nenhuma enquanto n?
|
|
398
|
+
?rica como o ant?doto capaz de dar sentido a o seu enorme poder .
|
|
399
|
+
. . .
|
|
400
|
+
>>> concordance("vender") # doctest: +SKIP
|
|
401
|
+
er recebido uma encomenda para vender 4000 blindados a o Iraque .
|
|
402
|
+
m?rico_Amorim caso conseguisse vender o lote de ac??es de o empres?r
|
|
403
|
+
mpre ter jovens simp?ticos a ? vender ? chega ! }
|
|
404
|
+
Disse que o governo vai vender ? desde autom?vel at? particip
|
|
405
|
+
ndiciou ontem duas pessoas por vender carro com ?gio .
|
|
406
|
+
A inten??o de Fleury ? vender as a??es para equilibrar as fi
|
|
407
|
+
|
|
408
|
+
Part-of-Speech Tagging
|
|
409
|
+
----------------------
|
|
410
|
+
|
|
411
|
+
Let's begin by getting the tagged sentence data, and simplifying the tags
|
|
412
|
+
as described earlier.
|
|
413
|
+
|
|
414
|
+
>>> from nltk.corpus import floresta
|
|
415
|
+
>>> tsents = floresta.tagged_sents()
|
|
416
|
+
>>> tsents = [[(w.lower(),simplify_tag(t)) for (w,t) in sent] for sent in tsents if sent]
|
|
417
|
+
>>> train = tsents[100:]
|
|
418
|
+
>>> test = tsents[:100]
|
|
419
|
+
|
|
420
|
+
We already know that ``n`` is the most common tag, so we can set up a
|
|
421
|
+
default tagger that tags every word as a noun, and see how well it does:
|
|
422
|
+
|
|
423
|
+
>>> tagger0 = nltk.DefaultTagger('n')
|
|
424
|
+
>>> nltk.tag.accuracy(tagger0, test)
|
|
425
|
+
0.17697228144989338
|
|
426
|
+
|
|
427
|
+
Evidently, about one in every six words is a noun. Let's improve on this by
|
|
428
|
+
training a unigram tagger:
|
|
429
|
+
|
|
430
|
+
>>> tagger1 = nltk.UnigramTagger(train, backoff=tagger0)
|
|
431
|
+
>>> nltk.tag.accuracy(tagger1, test)
|
|
432
|
+
0.87029140014214645
|
|
433
|
+
|
|
434
|
+
Next a bigram tagger:
|
|
435
|
+
|
|
436
|
+
>>> tagger2 = nltk.BigramTagger(train, backoff=tagger1)
|
|
437
|
+
>>> nltk.tag.accuracy(tagger2, test)
|
|
438
|
+
0.89019189765458417
|
|
439
|
+
|
|
440
|
+
|
|
441
|
+
Sentence Segmentation
|
|
442
|
+
---------------------
|
|
443
|
+
|
|
444
|
+
Punkt is a language-neutral sentence segmentation tool. We
|
|
445
|
+
|
|
446
|
+
>>> sent_tokenizer=nltk.data.load('tokenizers/punkt/portuguese.pickle')
|
|
447
|
+
>>> raw_text = machado.raw('romance/marm05.txt')
|
|
448
|
+
>>> sentences = sent_tokenizer.tokenize(raw_text)
|
|
449
|
+
>>> for sent in sentences[1000:1005]:
|
|
450
|
+
... print("<<", sent, ">>")
|
|
451
|
+
...
|
|
452
|
+
<< Em verdade, parecia ainda mais mulher do que era;
|
|
453
|
+
seria criança nos seus folgares de moça; mas assim quieta, impassível, tinha a
|
|
454
|
+
compostura da mulher casada. >>
|
|
455
|
+
<< Talvez essa circunstância lhe diminuía um pouco da
|
|
456
|
+
graça virginal. >>
|
|
457
|
+
<< Depressa nos familiarizamos; a mãe fazia-lhe grandes elogios, eu
|
|
458
|
+
escutava-os de boa sombra, e ela sorria com os olhos fúlgidos, como se lá dentro
|
|
459
|
+
do cérebro lhe estivesse a voar uma borboletinha de asas de ouro e olhos de
|
|
460
|
+
diamante... >>
|
|
461
|
+
<< Digo lá dentro, porque cá fora o
|
|
462
|
+
que esvoaçou foi uma borboleta preta, que subitamente penetrou na varanda, e
|
|
463
|
+
começou a bater as asas em derredor de D. Eusébia. >>
|
|
464
|
+
<< D. Eusébia deu um grito,
|
|
465
|
+
levantou-se, praguejou umas palavras soltas: - T'esconjuro!... >>
|
|
466
|
+
|
|
467
|
+
The sentence tokenizer can be trained and evaluated on other text.
|
|
468
|
+
The source text (from the Floresta Portuguese Treebank) contains one sentence per line.
|
|
469
|
+
We read the text, split it into its lines, and then join these lines together using
|
|
470
|
+
spaces. Now the information about sentence breaks has been discarded. We split this
|
|
471
|
+
material into training and testing data:
|
|
472
|
+
|
|
473
|
+
>>> import os, nltk.test
|
|
474
|
+
>>> testdir = os.path.split(nltk.test.__file__)[0]
|
|
475
|
+
>>> text = open(os.path.join(testdir, 'floresta.txt'), 'rb').read().decode('ISO-8859-1')
|
|
476
|
+
>>> lines = text.split('\n')
|
|
477
|
+
>>> train = ' '.join(lines[10:])
|
|
478
|
+
>>> test = ' '.join(lines[:10])
|
|
479
|
+
|
|
480
|
+
Now we train the sentence segmenter (or sentence tokenizer) and use it on our test sentences:
|
|
481
|
+
|
|
482
|
+
>>> stok = nltk.PunktSentenceTokenizer(train)
|
|
483
|
+
>>> print(stok.tokenize(test))
|
|
484
|
+
['O 7 e Meio \xe9 um ex-libris da noite algarvia.',
|
|
485
|
+
'\xc9 uma das mais antigas discotecas do Algarve, situada em Albufeira,
|
|
486
|
+
que continua a manter os tra\xe7os decorativos e as clientelas de sempre.',
|
|
487
|
+
'\xc9 um pouco a vers\xe3o de uma esp\xe9cie de \xaboutro lado\xbb da noite,
|
|
488
|
+
a meio caminho entre os devaneios de uma fauna perif\xe9rica, seja de Lisboa,
|
|
489
|
+
Londres, Dublin ou Faro e Portim\xe3o, e a postura circunspecta dos fi\xe9is da casa,
|
|
490
|
+
que dela esperam a m\xfasica \xabgeracionista\xbb dos 60 ou dos 70.',
|
|
491
|
+
'N\xe3o deixa de ser, nos tempos que correm, um certo \xabvery typical\xbb algarvio,
|
|
492
|
+
cabe\xe7a de cartaz para os que querem fugir a algumas movimenta\xe7\xf5es nocturnas
|
|
493
|
+
j\xe1 a caminho da ritualiza\xe7\xe3o de massas, do g\xe9nero \xabvamos todos ao
|
|
494
|
+
Calypso e encontramo-nos na Locomia\xbb.',
|
|
495
|
+
'E assim, aos 2,5 milh\xf5es que o Minist\xe9rio do Planeamento e Administra\xe7\xe3o
|
|
496
|
+
do Territ\xf3rio j\xe1 gasta no pagamento do pessoal afecto a estes organismos,
|
|
497
|
+
v\xeam juntar-se os montantes das obras propriamente ditas, que os munic\xedpios,
|
|
498
|
+
j\xe1 com projectos na m\xe3o, v\xeam reivindicar junto do Executivo, como salienta
|
|
499
|
+
aquele membro do Governo.',
|
|
500
|
+
'E o dinheiro \xabn\xe3o falta s\xf3 \xe0s c\xe2maras\xbb, lembra o secret\xe1rio de Estado,
|
|
501
|
+
que considera que a solu\xe7\xe3o para as autarquias \xe9 \xabespecializarem-se em
|
|
502
|
+
fundos comunit\xe1rios\xbb.',
|
|
503
|
+
'Mas como, se muitas n\xe3o disp\xf5em, nos seus quadros, dos t\xe9cnicos necess\xe1rios?',
|
|
504
|
+
'\xabEncomendem-nos a projectistas de fora\xbb porque, se as obras vierem a ser financiadas,
|
|
505
|
+
eles at\xe9 saem de gra\xe7a, j\xe1 que, nesse caso, \xabos fundos comunit\xe1rios pagam
|
|
506
|
+
os projectos, o mesmo n\xe3o acontecendo quando eles s\xe3o feitos pelos GAT\xbb,
|
|
507
|
+
dado serem organismos do Estado.',
|
|
508
|
+
'Essa poder\xe1 vir a ser uma hip\xf3tese, at\xe9 porque, no terreno, a capacidade dos GAT
|
|
509
|
+
est\xe1 cada vez mais enfraquecida.',
|
|
510
|
+
'Alguns at\xe9 j\xe1 desapareceram, como o de Castro Verde, e outros t\xeam vindo a perder quadros.']
|
|
511
|
+
|
|
512
|
+
NLTK's data collection includes a trained model for Portuguese sentence
|
|
513
|
+
segmentation, which can be loaded as follows. It is faster to load a trained model than
|
|
514
|
+
to retrain it.
|
|
515
|
+
|
|
516
|
+
>>> stok = nltk.data.load('tokenizers/punkt/portuguese.pickle')
|
|
517
|
+
|
|
518
|
+
Stemming
|
|
519
|
+
--------
|
|
520
|
+
|
|
521
|
+
NLTK includes the RSLP Portuguese stemmer. Here we use it to stem some Portuguese text:
|
|
522
|
+
|
|
523
|
+
>>> stemmer = nltk.stem.RSLPStemmer()
|
|
524
|
+
>>> stemmer.stem("copiar")
|
|
525
|
+
'copi'
|
|
526
|
+
>>> stemmer.stem("paisagem")
|
|
527
|
+
'pais'
|
|
528
|
+
|
|
529
|
+
|
|
530
|
+
Stopwords
|
|
531
|
+
---------
|
|
532
|
+
|
|
533
|
+
NLTK includes Portuguese stopwords:
|
|
534
|
+
|
|
535
|
+
>>> stopwords = nltk.corpus.stopwords.words('portuguese')
|
|
536
|
+
>>> stopwords[:10]
|
|
537
|
+
['a', 'ao', 'aos', 'aquela', 'aquelas', 'aquele', 'aqueles', 'aquilo', 'as', 'at\xe9']
|
|
538
|
+
|
|
539
|
+
Now we can use these to filter text. Let's find the most frequent words (other than stopwords)
|
|
540
|
+
and print them in descending order of frequency:
|
|
541
|
+
|
|
542
|
+
>>> fd = nltk.FreqDist(w.lower() for w in floresta.words() if w not in stopwords)
|
|
543
|
+
>>> for word in list(fd.keys())[:20]:
|
|
544
|
+
... print(word, fd[word])
|
|
545
|
+
, 13444
|
|
546
|
+
. 7725
|
|
547
|
+
« 2369
|
|
548
|
+
» 2310
|
|
549
|
+
é 1305
|
|
550
|
+
o 1086
|
|
551
|
+
} 1047
|
|
552
|
+
{ 1044
|
|
553
|
+
a 897
|
|
554
|
+
; 633
|
|
555
|
+
em 516
|
|
556
|
+
ser 466
|
|
557
|
+
sobre 349
|
|
558
|
+
os 313
|
|
559
|
+
anos 301
|
|
560
|
+
ontem 292
|
|
561
|
+
ainda 279
|
|
562
|
+
segundo 256
|
|
563
|
+
ter 249
|
|
564
|
+
dois 231
|
|
565
|
+
|