Development of Bilingual Application Using Machine Transliteration:A Practical Case Study

This paper focuses on the use of transliteration approach for customizable localization support in small scale systems. Marathi, Devanagari based Indian language is considered for the customizable localization support with the use of machine transliteration and translation memory using phonemic based pure consonant approach. Marathi is one of the widely spoken languages in India especially in the state of Maharashtra. This work addresses the support of local language access to the user to input and retrieve the data in Marathi language on the fly, whereas the data is stored in database in default language, English. User can interact with the system in Marathi as well as in English. The designed middleware plays the role of transliteration, when user uses the local language Marathi. Middleware reads data from the database and transliterate it into Devanagari script Marathi and display it to user. The transliteration from English to Devanagari and vice versa is carried out with the help of translation memory.  This method solves the problem of extra space on the web server as well as complexity in web pages. This approach provides safe and cost effective method of localizing existing and new web pages stored on web server.