English word segmentation.
WordSegment is an Apache2 licensed module for English word segmentation, written in pure-Python, and based on a trillion-word corpus.
Data files are derived from the Google Web Trillion Word Corpus, as described by Thorsten Brants and Alex Franz, and distributed by the Linguistic Data Consortium. This module contains only a subset of that data. The unigram data includes only the most common 333,000 words. Similarly, bigram data includes only the most common 250,000 phrases. Every word and phrase is lowercased with punctuation removed.
Installing WordSegment is simple with pip:
$ pip install wordsegment
You can access documentation in the interpreter with Python’s built-in help function:
>>> import wordsegment >>> help(wordsegment)
In your own Python programs, you’ll mostly want to use segment to divide a phrase into a list of its parts:
>>> from wordsegment import segment >>> segment('thisisatest') ['this', 'is', 'a', 'test']
WordSegment also provides a command-line interface for batch processing. This interface accepts two arguments: in-file and out-file. Lines from in-file are iteratively segmented, joined by a space, and written to out-file. Input and output default to stdin and stdout respectively.
$ echo thisisatest | python -m wordsegment this is a test
The maximum segmented word length is 24 characters. Neither the unigram nor bigram data contain words exceeding that length. The corpus also excludes punctuation and all letters have been lowercased. Before segmenting text, clean is called to transform the input to a canonical form:
>>> from wordsegment import clean >>> clean('She said, "Python rocks!"') 'shesaidpythonrocks' >>> segment('She said, "Python rocks!"') ['she', 'said', 'python', 'rocks']
Sometimes its interesting to explore the unigram and bigram counts themselves. These are stored in Python dictionaries mapping word to count.
>>> import wordsegment as ws >>> ws.load() >>> ws.UNIGRAMS['the'] 23135851162.0 >>> ws.UNIGRAMS['gray'] 21424658.0 >>> ws.UNIGRAMS['grey'] 18276942.0
Above we see that the spelling gray is more common than the spelling grey.
Bigrams are joined by a space:
>>> import heapq >>> from pprint import pprint >>> from operator import itemgetter >>> pprint(heapq.nlargest(10, ws.BIGRAMS.items(), itemgetter(1))) [('of the', 2766332391.0), ('in the', 1628795324.0), ('to the', 1139248999.0), ('on the', 800328815.0), ('for the', 692874802.0), ('and the', 629726893.0), ('to be', 505148997.0), ('is a', 476718990.0), ('with the', 461331348.0), ('from the', 428303219.0)]
Some bigrams begin with <s>. This is to indicate the start of a bigram:
>>> ws.BIGRAMS['<s> where'] 15419048.0 >>> ws.BIGRAMS['<s> what'] 11779290.0
The unigrams and bigrams data is stored in the wordsegment_data directory in the unigrams.txt and bigrams.txt files respectively.
Copyright 2016 Grant Jenks
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in compliance with the License. You may obtain a copy of the License at
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|File Name & Checksum SHA256 Checksum Help||Version||File Type||Upload Date|
|wordsegment-0.8.0-py2.py3-none-any.whl (4.4 MB) Copy SHA256 Checksum SHA256||py2.py3||Wheel||Jan 13, 2017|
|wordsegment-0.8.0.tar.gz (4.4 MB) Copy SHA256 Checksum SHA256||–||Source||Jan 13, 2017|