Influence of Contextual Information on Bengali-English Forward and Backward Transliteration Using Binary Coding

Transliteration of words across different scripts carries importance for both social media platforms and messaging apps. It preserves the phonetic meaning of words across various writing systems. This study evaluates the influence of contextual information within words on the transliteration of Bengali and English language pair. It utilizes the Dakshina and NEWS 2018 datasets and binary encoding for feature representation. With the random forest classifier, the results demonstrate that using neighboring transliteration units as context yields the best performance on both datasets, obtaining the highest accuracy of 81.04%. These findings highlight the vital role of contextual information within words in enhancing the performance of transliteration.