A Survey on Conversational AI Question answering System for Low Resource Language

Abstract

Question-and-answer (QA) systems allow users to pose questions about the information accessible in a variety of formats, such as structured and unstructured information in natural languages. It contributes significantly to conversational artificial intelligence (AI), which has led to the creation of a novel research field on conversational question answering (CQA), in which a system is required to understand the provided context before engaging in multi-turn QA to satisfy a user's informational needs. While the majority of currently done research focuses on single-turn QA, pre-trained language models and the availability of large multi-turn QA datasets have recently made multi-turn QA more significant. This study aims to give a thorough analysis of the most recent CQA research trends, mostly based on publications that have been reviewed in recent years. Our data indicates a shift in the industry's preference for multi-turn QA over single-turn QA, which has numerous advantages for conversational AI. In order to create a solid basis for the field of CQA, this survey aims to serve as the pinnacle for the research community.