CorrectMe

A context based auto-correction python package

CorrectMe is a context based auto-correction python package, also available with a GUI for demo purposes. It recommends correct words for misspellings by learning from the context.

All contributions are welcomed :)

Installation

Please use Python3.x for working with the package.

Creating a virtual environment is recommended as the code is in pre-release. Below steps are shown for conda environment.

conda create -n cm_env python=3.6 pip
source activate cm_env # activate cm_env for windows
cd CorrectMe # The cloned repository directory
pip install .

Working with the package

import correctme

Then you can call any of the correctme functions.

NOTE: API docs are not present yet. Contributing to docs are welcome. (Issue #27)

Working with the GUI

Type the below command in the terminal:

correctme-gui

This will pop up the GUI for the project, and you would also see the predictions in the terminal.

Current status of the project

Completed

• Performs distance based and phoneme based corrections.
• Installable Python library.
• Workable Kivy GUI.
• Almost complete test suite in pytest

WIP

• Still doesn't perform context based correction (Issue #20)
• Lacks documentation API docs, and user documentation. (Issue #27)

Algorithms used in the project

The idea is to make a context based correction package, where the package would perform auto-corrections, provide suggestions for corrections as well.

• Currently used algorithms:
  • bk_tree for finding closest words using Minimum Edit Distance.
  • double_metaphone to get closest words according to phonetics.