Urdu Spell Checking: Reverse Edit Distance Approach
Saadat Iqbal, Muhammad Waqas Anwar, Usama Ijaz Bajwa, Zobia Rehman · 2013
Spelling of words of a language are standardized by language authorities or consortiums and available in dictionaries or lexicons. For instance, “produkt ” does not belong to English dictionary. Similarly “نايمرد “ is a correctly spelled word, while “رنايمر د “ is a non-word in Urdu. Electronic representation of text is commonly used in today’s computing environment. The rich resourced languages like English have many applications with added tools. On the other hand, application development is in its infancy for less resourced language like Urdu. Spelling plays a vital role while humans write text electronically in computers. It is oblivious that terabytes of text is added in form of corpus or otherwise that is required to be spell checked, which is practically impossible to be done manually. In this work, various techniques for spellchecking have been studied and analyzed. All of them separately or a combination thereof can be used for the process of spell checking. Edit Distance technique has been widely used in spell checkers of various language, and a variation of this technique i.e. Reverse Edit Distance technique selected for suggesting correct words for nonwords. For Urdu, candidates are found by making 86n+41 comparisons for an ’n character ’ length Urdu word. 1