Abstract

A proof of concept Artificial Intelligence (AI) based application has been developed for a Question Answering system based on documents from Geological and Geophysical (G & G) domain by using Natural Language Processing (NLP) techniques. The present paper discusses the complete workflow of the Question Answering system, starting from data annotation, using pretrained BERT model, training/fine tuning the BERT model with new annotated dataset, preparing the data corpus and finally developing a web based user interface to access the core application.