Transliteration is a process in which input in source language is converted to an output in target language 1.
In another definition, Transliteration is a process of changes the words of source language to target language. To translate named entities across language, we commonly used the transliteration. In spelling correction, Information retrieving and extraction, Machine translation etc., transliteration is used. For those languages which have different writing or sound system, transliteration process is also difficult. But with the help deep learning all these issues will be easily removed and will increase the efficiency and accuracy of the existing system. In Transliteration, when map the source language string into target language string, then only script of that string is changed.
Transliteration process can be divided into two parts:- In first part segmentation of the source language string into small units of transliteration and in the second part relate these small units of transliteration with the target language string and also resolve the some conventions which are occur during the mapping of source to target language. When we want to convert or translate the proper names like Inderjeet or technical terms from Roman to non-Roman language, then transliteration is very useful for this type of translation. For those languages which have sound and writing system is different, Transliteration may define complex conventions and make that sure it perform the perfect or more accurate mapping from source to target language. For these type of languages, preserved the syllable sound of word by Transliteration process. There are two typed namely Transliteration, first one is Forward Transliteration and second one is Backward Transliteration. When we perform the transliteration from native language to non-native language then it is called Forward Transliteration. When restore the word from non-native language to native language then it is called Backward Transliteration.
To perform the Transliteration, there are many techniques or methods are used. But in general Transliteration model, there are two stages: In the first stage, train the data on a corpus and in the second stage, perform the Transliteration. In the training stage, words of source-target language trained at character or sound level. And in the transliteration, take the new word of source language and convert into appropriate target language word.