A Review on BERT and Its Implementation in Various NLP Tasks
V. A. Chakkarwar, Sharvari Tamane, Ankita Thombre
Abstract
We present a detailed study of BERT, which stands for 'Bidirectional Encoder Representations from Transformers'.Natural Language Processing can be considered as an important field when concerned with development of intelligent systems.Various tasks require understanding the correct meaning of the sentence in order to provide the output.Languages are difficult to understand by computers due to their ever-changing nature with context.BERT is considered as a revolution for making the computers understand the context of the text, which is the biggest hurdle in natural language processing tasks.It learns the language and its context in a way that closely resembles how a human brain understand the meaning of a sentence.It is unique because of its ability to learn from both left and right context in a sentence for a specific word.The evolution of BERT has marked a new era in perception and understanding of natural languages which may lead computers to grasps the natural language with better comprehension.The purpose of this paper is in the effort of providing a better understanding of BERT language model and its implementation in various NLP tasks.