Question answering model construction by using transfer learning

Every day there is a need for querying different kinds of data for a number of purposes and tasks. Therefore it became very important to have a tool like a Question Answering System (QAS) for information retrieving from any form of data in a quick manner. In this work a QAS which combines major fields of modern research topics: Deep Learning (DL), Natural Language Processing (NLP) and Information Technology (IT) is developed. Primary goal of this system is to provide users with an appropriate answer, based on the information that it has. Also an implementation of transfer learning on pre-trained bidirctionl nodr for question answering tasks is researched. Developed model is focused on question answering and was trained on a special Stanford dataset for QAS. The work explains the model’s structure, workflow and implementation in a detailed manner.