How To Create A Spell Checker In Python

pyspellchecker

Pure Python Spell Checking based on Peter Norvig's blog post on setting up a simple spell checking algorithm.

It uses a Levenshtein Distance algorithm to find permutations within an edit distance of 2 from the original word. It then compares all permutations (insertions, deletions, replacements, and transpositions) to known words in a word frequency list. Those words that are found more often in the frequency list are more likely the correct results.

pyspellchecker supports multiple languages including English, Spanish, German, French, and Portuguese. For information on how the dictionaries were created and how they can be updated and improved, please see the Dictionary Creation and Updating section of the readme!

pyspellchecker supports Python 3

pyspellchecker allows for the setting of the Levenshtein Distance (up to two) to check. For longer words, it is highly recommended to use a distance of 1 and not the default 2. See the quickstart to find how one can change the distance parameter.

Installation

The easiest method to install is using pip:

pip install pyspellchecker

To install from source:

git clone https://github.com/barrust/pyspellchecker.git              cd              pyspellchecker python setup.py install

For python 2.7 support, install release 0.5.6 but note that no future updates will support python 2.

pip install pyspellchecker==0.5.6

Quickstart

After installation, using pyspellchecker should be fairly straight forward:

              from              spellchecker              import              SpellChecker              spell              =              SpellChecker()              # find those words that may be misspelled              misspelled              =              spell.unknown(['something',              'is',              'hapenning',              'here'])              for              word              in              misspelled:              # Get the one `most likely` answer              print(spell.correction(word))              # Get a list of `likely` options              print(spell.candidates(word))

If the Word Frequency list is not to your liking, you can add additional text to generate a more appropriate list for your use case.

              from              spellchecker              import              SpellChecker              spell              =              SpellChecker()              # loads default word frequency list              spell.word_frequency.load_text_file('./my_free_text_doc.txt')              # if I just want to make sure some words are not flagged as misspelled              spell.word_frequency.load_words(['microsoft',              'apple',              'google'])              spell.known(['microsoft',              'google'])              # will return both now!

If the words that you wish to check are long, it is recommended to reduce the distance to 1. This can be accomplished either when initializing the spell check class or after the fact.

              from              spellchecker              import              SpellChecker              spell              =              SpellChecker(distance              =              1)              # set at initialization              # do some work on longer words              spell.distance              =              2              # set the distance parameter back to the default

Non-English Dictionaries

pyspellchecker supports several default dictionaries as part of the default package. Each is simple to use when initializing the dictionary:

              from              spellchecker              import              SpellChecker              english              =              SpellChecker()              # the default is English (language='en')              spanish              =              SpellChecker(language              =              'es')              # use the Spanish Dictionary              russian              =              SpellChecker(language              =              'ru')              # use the Russian Dictionary

The currently supported dictionaries are:

English - 'en'
Spanish - 'es'
French - 'fr'
Portuguese - 'pt'
German - 'de'
Russian - 'ru'

Dictionary Creation and Updating

The creation of the dictionaries is, unfortunately, not an exact science. I have provided a script that, given a text file of sentences (in this case from OpenSubtitles) it will generate a word frequency list based on the words found within the text. The script then attempts to *clean up* the word frequency by, for example, removing words with invalid characters (usually from other languages), removing low count terms (misspellings?) and attempts to enforce rules as available (no more than one accent per word in Spanish). Then it removes words from a list of known words that are to be removed. It then adds words into the dictionary that are known to be missing or were removed for being too low frequency.

The script can be found here: scripts/build_dictionary.py`. The original word frequency list parsed from OpenSubtitles can be found in the `scripts/data/` folder along with each language's include and exclude text files.

Any help in updating and maintaining the dictionaries would be greatly desired. To do this, a discussion could be started on GitHub or pull requests to update the include and exclude files could be added.

Additional Methods

On-line documentation is available; below contains the cliff-notes version of some of the available functions:

correction(word): Returns the most probable result for the misspelled word

candidates(word): Returns a set of possible candidates for the misspelled word

known([words]): Returns those words that are in the word frequency list

unknown([words]): Returns those words that are not in the frequency list

word_probability(word): The frequency of the given word out of all words in the frequency list

The following are less likely to be needed by the user but are available:

edit_distance_1(word): Returns a set of all strings at a Levenshtein Distance of one based on the alphabet of the selected language

edit_distance_2(word): Returns a set of all strings at a Levenshtein Distance of two based on the alphabet of the selected language

Credits

Peter Norvig blog post on setting up a simple spell checking algorithm
P Lison and J Tiedemann, 2016, OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles. In Proceedings of the 10th International Conference on Language Resources and Evaluation (LREC 2016)