unipy_nlp package

Subpackages

• unipy_nlp.analyze package
  • Submodules
  • unipy_nlp.analyze.topic_modeling module
  • unipy_nlp.analyze.word2vec module
  • Module contents
• unipy_nlp.test package
  • Submodules
  • unipy_nlp.test.hunspell_desc module
  • unipy_nlp.test.test_analyzer module
  • unipy_nlp.test.test_data_collector module
  • unipy_nlp.test.test_network_plot module
  • unipy_nlp.test.test_preprocessing module
  • unipy_nlp.test.test_tagger module
  • unipy_nlp.test.test_topic_modeling module
  • unipy_nlp.test.test_word2vec module
  • Module contents

Submodules

unipy_nlp.data_collector module

Get Data from xlsx.

unipy_nlp.data_collector.load_from_excel(filepath) → collections.abc.Generator

unipy_nlp.data_collector.recursive_substitutor(str_or_series, pattern_list, flags=0)

unipy_nlp.data_collector.refine_nested_excel_to_dict(xlsx_loaded) → pandas.core.frame.DataFrame

unipy_nlp.data_collector.collect_data(filepath, dump_filepath=None, dump_json_ok=True, return_tuple=True)

Summary

This function is for to gather text from xslx rawdata. Not designed for general uses.

Parameters

• filepath (str) – A directory xslx file(s) exists.

• dump_json_ok (bool (default: True)) – True if

• how ({'equal', 'remaining'}) – The method to split. ‘equal’ is to split chunks with the approximate length within the given size. ‘remaining’ is to split chunks with the given size, and the remains are bound as the last chunk.

• size (int) – The number of chunks.

Returns

A list of chunks.

Return type

list

Examples

>>> up.splitter(list(range(10)), how='equal', size=3)
[(0, 1, 2, 3), (4, 5, 6), (7, 8, 9)]

>>> up.splitter(list(range(10)), how='remaining', size=3)
[(0, 1, 2), (3, 4, 5), (6, 7, 8), (9,)]

unipy_nlp.network_plot module

A N-gram network plot.

class unipy_nlp.network_plot.WordNetwork(topic_freq_df, top_relevant_terms_df)

Bases: object

A network plot of co-occurance of words.

Parameters

• topic_freq_df (list) – A rank table by topic frequency.

• top_relevant_terms_df (list) – A rank table of Category.

pyvis_net

Type

pyvis.network.Network

ngramed_list

Type

list

ngramed_df

Type

pandas.DataFrame

get_ngram()

save_ngram()

load_ngram()

draw()

get_topic_mutuality_score_dict()

get_network_scored_repr_docs()

See also

Preprocessing

unipy_nlp.preprocessing.Preprocessor

Topic

unipy_nlp.analyze.topic_modeling.Topic_modeler

POS-Tagging

konlpy.tag.Mecab

Byte-Pair

sentencepiece

Examples

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.save_ngram('data/_tmp_dump/network_plot/ngram.json', type='json')
>>> vnet.save_ngram('data/_tmp_dump/network_plot/ngram.csv', type='csv')
>>> vnet.load_ngram('data/_tmp_dump/network_plot/ngram.json', type='json')
>>> vnet.load_ngram('data/_tmp_dump/network_plot/ngram.csv', type='csv')
>>> vnet.draw(
...     height="100%",
...     width='800px',
...     bgcolor='#ffffff',
...     font_color='black',
...     directed=True,
...     topic_top_n=5,
...     node_freq_threshold=100,
...     show_buttons=True,
... )
>>> (score_dict,
... score_dict_indiced) = vnet.get_topic_mutuality_score_dict(
...     cdict=tpm.corpora_dict
... )
>>> core_repr = vnet.get_network_scored_repr_docs(
...     bow_corpus=repr_bow_corpus_doc,
...     repr_docs=repr_sentenced,
...     save_ok=True,
...     savepath=None,
... )

draw(height='700px', width='800px', bgcolor='#ffffff', font_color='black', directed=True, notebook=False, topic_top_n=None, node_freq_threshold=None, show_buttons=True)

Draw pyvis.network.Network using N-grams.

Parameters

• height (str (default: “700px”)) –

  Height of the network plot. It can be pixel-based or percentage-based.

  width: str (default: “800px”)

  Height of the network plot. It can be pixel-based or percentage-based.

• bgcolor (str (default: ‘#ffffff’)) – HEX color for background.

• font_color (str (default: ‘black’)) – HEX color or colorname for font.

• directed (bool (default: True)) – An option to show direction for each edges.

• notebook (bool (default: False)) – An option to show in jupyter notebook

• topic_top_n (int (default: None)) – A topic number to show. It depends on its frequency.

• node_freq_threshold (int (default: None)) – A threshold number to show nodes. It is useful when your nodes & edges are too many to show.

• show_buttons (bool (default: True)) – An option to show interactive buttons in html.

Example

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.draw(
...     height="100%",
...     width='800px',
...     bgcolor='#ffffff',
...     font_color='black',
...     directed=True,
...     topic_top_n=5,
...     node_freq_threshold=100,
...     show_buttons=True,
... )

get_network_scored_repr_docs(bow_corpus, repr_docs, save_ok=True, filepath=None)

Get representitive documents, based on the mutuality score of terms.

Parameters

• bow_corpus (list) – A nested list, which contains converted documents into a list of token words.

• repr_docs (list) – A list of raw documents.

• save_ok (bool (default: True)) – An option to save.

• filepath (str (default: None)) – A filepath to save.

Examples

Example

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.draw(
...     height="100%",
...     width='800px',
...     bgcolor='#ffffff',
...     font_color='black',
...     directed=True,
...     topic_top_n=5,
...     node_freq_threshold=100,
...     show_buttons=True,
... )
>>> vnet.save('data/_tmp_dump/network_plot/vnet.html')
>>> (score_dict,
... score_dict_indiced) = vnet.get_topic_mutuality_score_dict(
...     cdict=tpm.corpora_dict
... )
>>> core_repr = vnet.get_network_scored_repr_docs(
...     bow_corpus=repr_bow_corpus_doc,
...     repr_docs=repr_sentenced,
...     save_ok=True,
...     savepath=None,
... )

get_ngram(tokenized_sentence_list)

Get N-grams for nodes & edges.

Parameters

tokenized_sentence_list (list) – A list of tokenized documents.

Examples

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)

get_topic_mutuality_score_dict(cdict)

Get scores of terms, based on its mutuality.

Parameters

cdict (gensim.corpora.dictionary.Dictionary) – A corpus dictionary for given documents.

Examples

Example

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.draw(
...     height="100%",
...     width='800px',
...     bgcolor='#ffffff',
...     font_color='black',
...     directed=True,
...     topic_top_n=5,
...     node_freq_threshold=100,
...     show_buttons=True,
... )
>>> vnet.save('data/_tmp_dump/network_plot/vnet.html')
>>> (score_dict,
... score_dict_indiced) = vnet.get_topic_mutuality_score_dict(
...     cdict=tpm.corpora_dict
... )

load_ngram(filename, type='json')

Load N-grams.

Parameters

• filepath (str) – A filepath to save.

• type (str (default: ‘json’, {‘json’, ‘csv’})) – Choose file type.

Examples

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.save_ngram('data/_tmp_dump/network_plot/ngram.json', type='json')
>>> vnet.load_ngram('data/_tmp_dump/network_plot/ngram.json', type='json')

save(filepath_html)

Save pyvis.network.Network.

Parameters

filepath_html (str) – A filepath to save.

Examples

Example

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.draw(
...     height="100%",
...     width='800px',
...     bgcolor='#ffffff',
...     font_color='black',
...     directed=True,
...     topic_top_n=5,
...     node_freq_threshold=100,
...     show_buttons=True,
... )
>>> vnet.save('data/_tmp_dump/network_plot/vnet.html')

save_ngram(filepath, type='json')

Save N-grams.

Parameters

• filepath (str) – A filepath to save.

• type (str (default: ‘json’, {‘json’, ‘csv’})) – Choose file type.

Examples

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> import unipy_nlp.analyze.topic_modeling as utpm
>>> import unipy_nlp.network_plot as unet
>>> tpm = utpm.TopicModeler(sentence_list, tokenized)
>>> tpm.train_lda(...)
>>> tpm.visualize_lda_to_html(...)
>>> vnet = unet.WordNetwork(
...     topic_freq_df=tpm.topic_freq_df,
...     top_relevant_terms_df=tpm.top_relevant_terms_df,
... )
>>> vnet.get_ngram(tokenized)
>>> vnet.save_ngram('data/_tmp_dump/network_plot/ngram.json', type='json')

unipy_nlp.network_plot.feature_map(arr)

unipy_nlp.network_plot.sg_pair_array(line_token_ids, window_size=5, num_grams=6)

unipy_nlp.network_plot.sg_pair_list(line_token_ids, window_size=5, num_grams=6)

unipy_nlp.network_plot.skip_gram_pairs(token_ids, window_size=2, num_grams=4, shuffle=False)

unipy_nlp.network_plot.token_pair_extractor(token_ls)

unipy_nlp.preprocessing module

Tokenize text with sentencepiece & MeCab, from xlsx & Elasticsearch.

class unipy_nlp.preprocessing.Preprocessor(tagger='mecab')

Bases: object

Text Preprocessing with POS-Tagging or Byte-Pair Encoding.

Get tokenized from text.

Parameters

tagger (str {‘mecab’,}) – A POS-Tagging Engine to use.

source_sentences

sentences from text, given by read_json or read_es.

Type

list

read_json()

read_es()

pos_tag()

train_spm()

load_spm()

spm_encode()

See also

POS-Tagging

konlpy.tag.Mecab

Byte-Pair

sentencepiece

Examples

>>> import unipy_nlp.data_collector as udcl
>>> import unipy_nlp.preprocessing as uprc
>>> from pprint import pprint
>>> prep = uprc.Preprocessor()
>>> prep.read_json('./data/_tmp_dump/prep/rawdata_collected.json')
>>> sentence_for_pos_list = [
...     "무궁화 꽃이 피었습니다."
...     "우리는 민족중흥의 역사적 사명을 띠고 이 땅에 태어났다.",
... ]
>>> tokenized_morphed_filtered = prep.pos_tag(
...     input_text=sentence_for_pos_list,
...     tag_type=[
...         '체언 접두사', '명사', '한자', '외국어',
...         '수사', '구분자',
...         '동사',
...         '부정 지정사', '긍정 지정사',
...     ]
... )
>>> print(tokenized_morphed_filtered)
[['무궁화'], ['우리', '민족중흥', '역사', '사명']]
>>> prep.train_spm(
...     source_type='list',
...     model_type='bpe',
...     vocab_size=30000,
...     model_name='spm_trained',
...     savepath='./data/_tmp_dump/spmed',
...     random_seed=1,
... )
>>> prep.load_spm(
...     savepath='./data/_tmp_dump/spmed',
...     model_name='spm_trained',
...     use_bos=False,
...     use_eos=False,
...     vocab_min_freq_threshold=None,
... )
>>> sentence_for_spm_list = [
...     "새로운 기술환경의 발전과 확산이 진행되는 it환경",
...     "비즈니스 환경과의 접목에 집중해 새로운 사업영역 선점",
... ]
>>> tokenized_spmed = prep.spm_encode(
...     sentence_for_spm_list,
...     type='piece',
...     rm_space=True,
... )
>>> pprint(tokenized_spmed)
[['새로운', '기술', '환경의', '발전과', '확산이', '진행되는', 'it', '환경'],
['비즈니스', '환경', '과의', '접목', '에', '집중', '해', '새로운', '사업영역', '선점'],

drop_by_minimum_length(sentence_list, min=2)

load_spm(savepath='./data', model_name=None, use_bos=False, use_eos=False, vocab_min_freq_threshold=None)

A high-level wrapper for sentencepiece.SentencePieceTrainer.Load.

Parameters

• savepath (str (default: ‘./data’)) – A dirpath to load.

• model_name (str (default: ‘spm_trained’)) – A filename prefix to load.

• use_bos (bool (default: False)) – An option of SetEncodeExtraOptions.

• use_eos (bool (default: False)) – An option of SetEncodeExtraOptions.

• vocab_min_freq_threshold (int (default: None)) – An lower bound of vocabulary by its frequency.

Example

>>> import unipy_nlp.preprocessing as uprc
>>> prep = uprc.Preprocessor()
>>> prep.read_json('./data/_tmp_dump/prep/rawdata_collected.json')
>>> prep.load_spm(
...     savepath='./data/_tmp_dump/spmed',
...     model_name='spm_trained',
...     use_bos=False,
...     use_eos=False,
...     vocab_min_freq_threshold=None,
... )

pos_tag(input_text=None, tag_type=None)

POS-Tagging with input_text or pre-loaded sentences.

Parameters

• input_text (list (default: None)) – A list of sentences. If None, use self.source_sentences internally.

• tag_type (list (default: None)) – A tag name to subset. You can use ‘NNP’ or ‘일반 명사’ either.

Returns

tokenized

Return type

list

Example

>>> import unipy_nlp.preprocessing as uprc
>>> ES_HOST = '52.78.243.101'
>>> ES_PORT = '9200'
>>> prep = uprc.Preprocessor()
>>> sentence_for_pos_list = [
...     "무궁화 꽃이 피었습니다.",
...     "우리는 민족중흥의 역사적 사명을 띠고 이 땅에 태어났다.",
... ]
>>> tokenized = prep.pos_tag(
...     input_text=sentence_for_pos_list,
...     tag_type=[
...         '체언 접두사', '명사', '한자', '외국어',
...         '수사', '구분자',
...         '동사',
...         '부정 지정사', '긍정 지정사',
...         'NNP', 'NNG',
...     ]
... )
>>> print(tokenized)
[['무궁화'], ['우리', '민족중흥', '역사', '사명']]

read_es(host, port, index='happymap_temp', match_as_flat_dict=None, key='contents', drop_min=2)

Read sentences from Elasticsearch, as self.source_sentences.

Parameters

• host (str) – A domain address of Elasticsearch server.

• port (str) – A port number of Elasticsearch server.

• index (str) – An index of Elasticsearch server.

• match_as_flat_dict (str (default: None)) –

  An option to query_match. match_all If None. Example: ``` match_as_flat_dict={

  ’sheet_nm’: ‘2019’, ‘table_nm’: ‘board’,

• key (str) – A key of sentences in an object.

• drop_min (int (default: 2)) – A lower bond of sentence length. If an inappropriate value is given, it will be changed by 1 systemically.

Example

>>> import unipy_nlp.preprocessing as uprc
>>> ES_HOST = '52.78.243.101'
>>> ES_PORT = '9200'
>>> prep = uprc.Preprocessor()
>>> prep.read_es(
...     host=ES_HOST,
...     port=ES_PORT,
...     index='logs',
...     match_as_flat_dict={
...         'sheet_nm': '2019',
...         'table_nm': 'board',
...     },
...     key='contents',
...     drop_min=2,
... )
>>> prep.source_sentences[:2]
['새로운 기술환경의 발전과 확산이 진행되는 it환경', '비즈니스 환경과의 접목에 집중해 새로운 사업영역 선점']

read_json(filename, key='contents', drop_min=2)

Read sentences from disk, as self.source_sentences.

Parameters

• filename (str) – A filepath to read.

• key (str) – A key of sentences in json object.

• drop_min (int (default: 2)) – A lower bond of sentence length. If an inappropriate value is given, it will be changed by 1 systemically.

Example

>>> import unipy_nlp.preprocessing as uprc
>>> prep = uprc.Preprocessor()
>>> prep.read_json(
...     './data/_tmp_dump/prep/rawdata_collected.json',
...     key='contents',
...     drop_min=2,
... )
>>> prep.source_sentences[:2]
['새로운 기술환경의 발전과 확산이 진행되는 it환경', '비즈니스 환경과의 접목에 집중해 새로운 사업영역 선점']

spm_encode(input_list, type='piece', rm_space=True)

A high-level wrapper for sentencepiece.EncodeAsPieces or sentencepiece.EncodeAsIds.

Parameters

• input_list (list) – A list of sentences to tokenize.

• type (str (default: ‘piece’, {‘piece’, ‘id’})) – Choose encoding type. ‘piece’: str, ‘id’: int

• rm_space (bool (default: True)) – An option to remove “▁” (U+2581), which represents the whitespace.

Example

>>> import unipy_nlp.preprocessing as uprc
>>> prep = uprc.Preprocessor()
>>> prep.read_json('./data/_tmp_dump/prep/rawdata_collected.json')
>>> prep.load_spm(
...     savepath='./data/_tmp_dump/spmed',
...     model_name='spm_trained',
...     use_bos=False,
...     use_eos=False,
...     vocab_min_freq_threshold=None,
... )
>>> sentence_for_spm_list = [
...     "새로운 기술환경의 발전과 확산이 진행되는 it환경",
...     "비즈니스 환경과의 접목에 집중해 새로운 사업영역 선점",
... ]
>>> tokenized_spmed = prep.spm_encode(
...     sentence_for_spm_list,
...     type='piece',
...     rm_space=True,
... )
>>> pprint(tokenized_spmed)
[['새로운', '기술', '환경의', '발전과', '확산이', '진행되는', 'it', '환경'],
['비즈니스', '환경', '과의', '접목', '에', '집중', '해', '새로운', '사업영역', '선점'],

train_spm(source_type='list', source_file=None, model_type='bpe', vocab_size=50000, model_name='spm_trained', savepath='./data', random_seed=None)

A high-level wrapper for sentencepiece.SentencePieceTrainer.Train.

Parameters

• source_type (str (default: ‘list’)) – list: Use self.source_sentences as an input. txt: Use a given text file as an input. It should be split by a sentence.

• model_type (str (default: ‘bpe’, {‘bpe’, ‘word’, ‘char’, ‘unigram’})) – A model_type of sentencepiece.

• vocab_size (int (default: 50000)) – Embedding size of sentencepiece.

• model_name (str (default: ‘spm_trained’)) – A filename prefix to save.

• savepath (str (default: ‘./data’)) – A dirpath to save.

• random_seed (int (default: None)) – A random seed number.

Example

>>> import unipy_nlp.preprocessing as uprc
>>> prep = uprc.Preprocessor()
>>> prep.read_json('./data/_tmp_dump/prep/rawdata_collected.json')
>>> prep.train_spm(
...     source_type='list',
...     model_type='bpe',
...     vocab_size=30000,
...     model_name='spm_trained',
...     savepath='./data/_tmp_dump/spmed',
...     random_seed=1,
... )

unipy_nlp.tagger module

class unipy_nlp.tagger.Mecab(dicpath='/home/docs/checkouts/readthedocs.org/user_builds/unipy-nlp/checkouts/latest/unipy_nlp/_resources/mecab/mecab/lib/mecab/dic/mecab-ko-dic')

Bases: object

Wrapper for MeCab-ko morphological analyzer.

MeCab, originally a Japanese morphological analyzer and POS tagger developed by the Graduate School of Informatics in Kyoto University, was modified to MeCab-ko by the Eunjeon Project to adapt to the Korean language.

In order to use MeCab-ko within KoNLPy, follow the directions in optional-installations.

Parameters

dicpath – The path of the MeCab-ko dictionary.

morphs(phrase)

Parse phrase to morphemes.

nouns(phrase)

Noun extractor.

pos(phrase, flatten=True, join=False)

POS tagger.

Parameters

• flatten – If False, preserves eojeols.

• join – If True, returns joined sets of morph and tag.

unipy_nlp.tagger.build_mecab_user_dic(nested_list, mode='a')

Module contents

NLP Analysis Tools.

unipy_nlp

Provides

1. NLP Data Handling Tools

2. NLP Analysis Functions.

3. POS-Tagger (MeCab, non-sudo intall required.)

4. Topic Modeling(LDA)

5. Word2Vec

6. NLP network plot.

How to use

In terms of Data science, Data Preprocessing & Plotting is one of the most annoying parts of Data Analysis. unipy offers you many functions maybe once you have tried to search in google or stackoverflow.

The docstring examples assume that unipy has been imported as up::

>>> import unipy_nlp as unlp

Use the built-in help function to view a function’s docstring::

>>> help(np.sort)
... # doctest: +SKIP

General-purpose documents like a glossary and help on the basic concepts of numpy are available under the docs sub-module:

>>> from unipy import docs
>>> help(docs)
... 

Available subpackages

data_collector

Get Data from xlsx.

tagger

A MeCab Wrapper without installation.

preprocessing

Tokenize text with sentencepiece & MeCab, from xlsx & Elasticsearch.

analyze

Topic Modeling(LDA) & Word2Vec.

network_plot

A N-gram network plot.

test

Test-codes of unipy_nlp

This module provides a number of useful functions for natural language handling.

analyze

• func – Description here.

unipy_nlp.build_mecab()