/spellchecker

Spell correct entire sentences using nltk freqdist and symspell

Primary LanguagePython

Ola Spellchecker

Reference for Word probability http://www.katrinerk.com/courses/python-worksheets/language-models-in-python

  1. Need to match word case (DONE)
  2. Preserve spaces (DONE)
  3. Word not in corpus wont be spell corrected
  4. Handle multiple spelling mistakes in one sentence Ther is nthing wrng (DONE)

Questions

Following https://www.microsoft.com/cognitive-services/en-us/bing-spell-check-api

  1. To handle broken words (microso ft) => Use this https://github.com/grantjenks/wordsegment ?
  2. Slang => Synonyms
  3. Names => Our lookup store
  4. Homonyms => Supported
  5. Brands => Our lookup store

Usage

pip install -e git+ssh://git@github.com/rmdort/ola_spellchecker.git#egg=ola_spellchecker

Requirements

  1. Python
  2. NLTK

Usage

from ola_language_tools import SpellCheck

# Create an instance of the class
spellchecker = SpellCheck(corpus='spellcheck-corpus.txt')

print spellchecker.correct('Wher is everone?')

How it works

  1. Create a dictionary out of a large corpus of text
  2. Identify spell mistake for each word in the query
  3. Find the probability of co-occurence of each mis-spelled word correction
  4. Use the best probable match as replacement