I require a machine learning model to be used in hardware keyboard testing.
It needs to work in reverse of spelling correction. For example given a word, based on an existing dataset, the output should be a list of misspellings of this word in a sorted order of most likely to least likely (with a variable limit)
So in summary:
To use dataset consisting of word and misspelled version of the word per row.
input = word
output = array of misspelled words
This will be used in testing datasets of various custom made hardware keyboards, and optimising those to minimise certain common misspellings.
No language requirements, but needs to run on linux and be able to accept command line arguments. Also should be easy to deploy on ubuntu linux nodes, like Amazon AWS.