TextRank4ZH Documentation

You can install TextRank4ZH using pip or by running the setup.py script. For Python 3, ensure you use python3 and pip3 commands.

Using pip (Recommended):

# Install for current user
pip3 install textrank4zh --user

# System-wide install
sudo pip3 install textrank4zh

Using setup.py:

# Install for current user
python3 setup.py install --user

# System-wide install
sudo python3 setup.py install

To remove the package from your environment, use the pip uninstall command.

pip uninstall textrank4zh

Use the TextRank4Sentence class to extract the most important sentences from a text to create a summary.

Methods:

• analyze(text, lower=True, source='all_filters'): Processes the input text.
• get_key_sentences(num=3): Returns a list of sentence objects. Each object contains:
  • .index: The position of the sentence in the original text.
  • .weight: The importance weight of the sentence.
  • .sentence: The actual sentence text.

Use the TextRank4Keyword class to extract important words and adjacent keyword groups (keyphrases) from Chinese text.

Methods:

• analyze(text, lower=True, window=2): Processes the input text.
  • In Python 2, text must be a str (UTF-8) or unicode object.
  • In Python 3, text must be a bytes or str object.
• get_keywords(num, word_min_len=1): Returns a list of keyword objects containing .word and .weight.
• get_keyphrases(keywords_num=20, min_occur_num=2): Returns a list of keyphrase strings.

from textrank4zh import TextRank4Keyword

text = "你的中文文本"
tr4w = TextRank4Keyword()
tr4w.analyze(text=text, lower=True, window=2)

# Get keywords
for item in tr4w.get_keywords(20, word_min_len=1):
    print(item.word, item.weight)

# Get keyphrases
for phrase in tr4w.get_keyphrases(keywords_num=20, min_occur_num=2):
    print(phrase)

Both TextRank4Keyword and TextRank4Sentence break down the input text into four internal formats after calling .analyze():

1. sentences: A list of sentences.
2. words_no_filter: A 2D list of words (tokenized sentences) without any filtering.
3. words_no_stop_words: A 2D list of words after removing stop words.
4. words_all_filters: A 2D list of words after removing stop words and filtering by specific parts of speech (POS).

TextRank4ZH requires the following Python packages:

• jieba >= 0.35
• numpy >= 1.7.1
• networkx >= 1.9.1

Compatibility: Tested successfully on Python 2.7.9 and Python 3.4.3.