An Approach Towards Multilingual Translation By Semantic-Based Verb Identification And Root Word Analysis
Md. Saidul Hoque Anik, Md. Adnanul Islam, A. B. M. Alim Al Islam · 2018
Popular and widely available translators like Google Translator uses statistic based approach to build the multilingual translation system. This approach solely depends on the availability of a large number of samples. Which is why, Google translator performs interestingly well when it translates among the popular languages like English, French or Spanish, however, makes elementary mistakes when it translates the languages that are newly introduced or less known to the system. Most of the research found so far on natural language processing (NLP), have been performed keeping English as the target language. However, a good number of widely spoken potential languages remain nearly unexplored in the research fields which is quite unexpected in the era of global communication. In this study, we have tried to explore a generalized machine translation system, especially for the languages having insufficient availability in literature. This study basically focuses on Bengali Language as an example of such low resource languages. In this work, we have proposed different approaches for semantic based verb identification along with its translation, and hence, developed an algorithm for root word detection of a verb in any sentence which reflects significant improvement over Google Translator. Finally, we have shown a comparison among the different approaches in terms of accuracy, time complexity and space complexity.