WNSpell: a WordNet-Based Spell Corrector

Bill X. Huang · 2016

This paper presents a standalone spell corrector, WNSpell, based on and written for WordNet.It is aimed at generating the best possible suggestion for a mistyped query but can also serve as an all-purpose spell corrector.The spell corrector consists of a standard initial correction system, which evaluates word entries using a multifaceted approach to achieve the best results, and a semantic recognition system, wherein given a related word input, the system will adjust the spelling suggestions accordingly.Both feature significant performance improvements over current context-free spell correctors.

Read the paper · More papers on PaperTik