A Comprehensive Review on Deep Learning Approaches for Question Answering and Machine Reading Comprehension in NLP

Natural Language Processing (NLP) deals with the development of methodologies capable of interacting with computer through human language. NLP improves machine’s comprehension of human language, allowing for human-computer communication based on linguistics. Recent years have seen phenomenal success of NLP models in language and grammatical tasks such as information extraction, translation, classification and reasoning. This accomplishment is mainly due to the influence of transformers, which inspired design ideas such as BERT, SQuAD 2.0 and others. These large-scale models produced unique results, despite the higher computational cost. As a result, current NLP systems use transfer learning, pruning, knowledge filtration and quantization to accomplish reasonable performance. Furthermore, Information Retrievers (IR) are created to extract precise data files from large datasets, addressing the large data assertion made by language models. Major contribution of this study is to understand the application of deep learning methods in NLP for automated question answering and obtaining a comprehension of essay text. Context-based NLP issues that are presented along with existing solutions. The challenges of using NLP in comprehension are examined, as well as research community methods for extracting answers from paragraphs. Further direction of this research is to develop novel deep learning models for QA and text comprehension that can overcome the demerits of existing approaches.