Evolution of Financial Question Answering Themes, Challenges, and Advances

Financial Question Answering (QA) has emerged as a critical area of research, aiming to develop intelligent systems capable of interpreting and answering complex queries within financial reports. This survey paper delves into the diverse landscape of Financial Question Answering (QA) research, grouping influential papers into thematic clusters to uncover varied perspectives and challenges within this evolving field through an in-depth analysis of the influential papers published over the decade. The survey begins by introducing the scope and significance of Financial QA, followed by a discussion on the evolution of datasets tailored for financial domain-specific QA tasks, focusing on conversational QA, numerical reasoning, and complex reasoning within financial reports. Subsequently, the survey delves into various model architectures and techniques employed in Financial QA. The exploration encompasses pivotal themes such as numerical reasoning and tabular data handling, conversational QA in finance, model architectures for financial QA, knowledge-infused QA, and datasets creation and introduction. These themes elucidate the challenges and possibilities inherent in Financial QA, emphasizing issues like numerical reasoning complexity, contextual understanding, hybrid data challenges, coreference, performance gaps, data limitations, and optimal context size. This survey offers a holistic view of the advancements, challenges, and future directions in Financial QA, fostering further innovation in this evolving field.