Data Driven Approach to Sinhala Spellchecker and Correction
L. G. B. Subhagya, L. Ranathunga, W. H. Anjanee Nimasha, B. R. Jayawickrama, K. L Mahaliyanaarchchi · 2018
This paper described the process of spell checker of Sinhala, that is major language of Sri Lanka. Due to similar structure and similar sounding letters of Sinhala language and misspelling, missing letters, caused for incorrect words. This approach is described based on permutation generation based on similar structure, similar sounding letters and minimum edit distance method. Then best suggestion is selected based on context-based analysis. Due to context-based analysis, it gives more accurate suggestion for incorrect word. This proposed method able to detect and correct incorrect words. The proposed solution is evaluated by precision, recall and f-measure based on one incorrect letter in word, two incorrect letters in word, missing modifiers and context-based approach evaluation methods. Overall method is shown average precision of 84.73%, recall of 83.69 % and f-measure of 86.01%.