Handling Inflections for Translation and Part of Speech (POS) Tagging of Bengali Words

Abstract

This paper presents a translation scheme and parsing strategy of Bengali words aiming to develop a machine translation (MT) system for Bangla to English. In this research, we have focused large diversity of the complex inflection variations in Bengali language. As the inflected words without exception are not readily available in the dictionary, identifying the root words or stems from inflected words are essential for correct translation. In our work, we decomposed the inflected words into its constituent parts namely root and suffix. We have proposed here an algorithm to perform this decomposition. Correct translation of each word and its morphological properties were identified with special types of dictionaries for suffixes along with the root word dictionary. To simplify the computer processing, a tag vector was used to represent the useful morphological properties. Dictionary entries for the root words were organized in alphabetical order, while the suffixes were stored in their corresponding dictionaries with logically choosing a hash function.

Cite this Article

Asaduzzaman A.O.M, Md. Rahman M. Handling Inflections for Translation and Part of Speech (POS) Tagging of Bengali Words. Journal of Computer Technology & Applications. 2016; 7(1): 27–34p.