A Survey on Machine Reading Comprehension Systems

Abstract Machine Reading Comprehension (MRC) is a challenging task and hot topic in Natural Language Processing. The goal of this field to develop systems for answering the questions regarding given context. In paper, we present comprehensive survey on diverse aspects MRC systems, including their approaches, structures, input/outputs, research novelties. We illustrate recent trends based review 241 papers published during 2016–2020. Our investigation demonstrated that focus has changed years from answer extraction generation, single- multi-document reading comprehension, learning scratch using pre-trained word vectors. Moreover, discuss popular datasets evaluation metrics field. paper ends with an most-cited contributions.