Detection and Correction of Homophonous Error Word for Khmer Language

by

Chea Sok Huor, Top Rithy, Ros Pich Hemy

Abstract

The ability to detect and correct the written error is a very crucial challenge for Khmer language computing because of the fact that Khmer does not separate words in its writing system. The richness of Khmer characters confuses users to write a word in many different ways. In this paper, we proposed a method to detect the homophonous non-word error using a Khmer Common Expression (in short KCE) and automatically correct it. The idea is to generate the same expression for every word that is likely to be confused in sound. Therefore, the string to map the word in the dictionary is not the real word string, but an expression, which has the same pronunciation as the target word. Download (PDF).