Financial Causality Detection

Abstract

This paper presents a Financial Causality Detection Software designed to identify causal effects in financial disclosures. Leveraging a hybrid approach of extractive and generative Question-answering (QA) models, the software processes financial documents to reveal causal relationships between variables and events. The system provides users with concise, context-based causal insights by utilizing state-of-the-art Natural Language Processing (NLP) techniques and Generative AI. The software supports multiple languages, focusing on English and Spanish datasets, making it a versatile tool for multilingual financial analysis. This paper explores the architecture, benefits, challenges, and implications of adopting such a system for financial professionals and regulatory bodies.