Identifying The Common Type of Spelling Error by Leveraging Levenshtein Distance and N-gram

Abstract

A spelling error commonly occurs during document writing. It probably happens due to the authors’ vocabulary incompetence or they may strike the improper key in the keyboard. The types of errors that mostly appear such as insertion of an extra letter, deletion of one letter, substitution of one letter, or transposition of two adjacent letters. This study aims to identify the common type of spelling error and it uses the list of common misspelling words submitted by Wikipedia contributors.A brief overview of Levenshtein and N-gram distance techniques is provided to describe the technical approaches that support the author to achieve the purpose of this study.Those two techniques are utilised to predict the correct word of misspellings from the English dictionary.This study shows that Levenshtein works well to correct substitution single letter and transposition two sequenced letters, while N-gram operates effectively to fix the word with letter omission.The overall result is then evaluated by recall measurement to see which technique that works well on correcting the misspellings. Since the recall of Levenshtein is higher than N-gram, it is concluded that the frequency of misspelling words which are correctly fixed by Levenshteinoccurs more often.