Sylvie THOUËSNY
School of Applied Language and Intercultural Studies, Dublin City University, Ireland
sylvie.thouesny@icall-research.net
|

|
A University of Pittsburgh study states that using the spellcheck button can create problems for those who rely too heavily on the software. Spelling checker
should just be a flag indicating that some errors could occur at
some stage, but certainly not a systematic way to correct text.
Even if the purpose of Computational Linguistics is to be very
close of human minds, we still have to think by ourselves.
This paper is an approach to present how probabilities can
be used in order to generate suggestions for misspelt words.
SpellChecker, as its name indicates, is a spell checker. It is
written in Java and is easy to use. It looks at single error and
generates a list of candidates for non-word errors.
Based on existing algorithms, SpellChecker detects
deletion, substitution, reversal or insertion errors according
to the dictionary in use. Candidates are computed according to
probabilities and sorted by the most likely. The user has the
possibility to replace the misspelt word suggested or simply
ignore it.
|
|