Use of a novel hash-table for speeding-up suggestions for misspelt Tamil words

Abstract

Spell checker is a tool that finds and corrects misspelt words in a text document. Spelling error detection and correction techniques are widely used by text editing systems, machine translation systems, optical character recognition systems, search engines and speech recognition systems. Though spell checkers for European languages and Indian languages are well developed, few for Tamil language, perhaps, because the fact that Tamil language is morphologically rich and agglutinative makes it a challenging task. An efficient approach to generating suggestions for misspelt words in Tamil language has been proposed in this paper. The proposed novel approach uses n-gram technique on stemmed form of the words with two different hash-tables and find the better one to generate most suitable alternatives to misspelt words by speeding up the lookup. The use of length of words in hash-table speed up finding appropriate suggestions while reducing the number of inappropriate suggestions. Test results show that the suggestions generated by the system are with 95% accuracy as approved by a Scholar in Tamil.