/wordninja

Probabilistically split concatenated words using NLP based on English Wikipedia unigram frequencies.

Primary LanguagePythonMIT LicenseMIT

image

Word Ninja

Slice your munged together words! Seriously, Take anything, 'imateapot' for example, would become ['im', 'a', 'teapot']. Useful for humanizing stuff (like database tables when people don't like underscores).

This project is repackaging the excellent work from here: http://stackoverflow.com/a/11642687/2449774

Usage

$ python
>>> import wordninja
>>> wordninja.split('derekanderson')
['derek', 'anderson']
>>> wordninja.split('imateapot')
['im', 'a', 'teapot']
>>> wordninja.split('heshotwhointhewhatnow')
['he', 'shot', 'who', 'in', 'the', 'what', 'now']
>>> wordninja.split('thequickbrownfoxjumpsoverthelazydog')
['the', 'quick', 'brown', 'fox', 'jumps', 'over', 'the', 'lazy', 'dog']

Performance

It's super fast!

>>> def f():
...   wordninja.split('imateapot')
... 
>>> timeit.timeit(f, number=10000)
0.40885152100236155

It can handle long strings:

>>> wordninja.split('wethepeopleoftheunitedstatesinordertoformamoreperfectunionestablishjusticeinsuredomestictranquilityprovideforthecommondefencepromotethegeneralwelfareandsecuretheblessingsoflibertytoourselvesandourposteritydoordainandestablishthisconstitutionfortheunitedstatesofamerica')
['we', 'the', 'people', 'of', 'the', 'united', 'states', 'in', 'order', 'to', 'form', 'a', 'more', 'perfect', 'union', 'establish', 'justice', 'in', 'sure', 'domestic', 'tranquility', 'provide', 'for', 'the', 'common', 'defence', 'promote', 'the', 'general', 'welfare', 'and', 'secure', 'the', 'blessings', 'of', 'liberty', 'to', 'ourselves', 'and', 'our', 'posterity', 'do', 'ordain', 'and', 'establish', 'this', 'constitution', 'for', 'the', 'united', 'states', 'of', 'america']

And scales well. (This string takes ~7ms to compute.)

How to Install

pip3 install wordninja

Custom Language Models

#1 most requested feature! If you want to do something other than english (or want to specify your own model of english), this is how you do it.

>>> lm = wordninja.LanguageModel('my_lang.txt.gz')
>>> lm.split('derek')
['der','ek']

Language files must be gziped text files with one word per line in decreasing order of probability.

If you want to make your model the default, set:

wordninja.DEFAULT_LANGUAGE_MODEL = wordninja.LanguageModel('my_lang.txt.gz')