Reasoning over Financial Scenarios with the Vadalog System

Abstract

The last decade has witnessed a surge of interest from the financial sector towards intelligent systems that model complex business domains via powerful languages and employ them to reason over enterprise data and infer new knowledge. Since many languages and systems for Knowledge Representation and Reasoning have recently appeared on the scene, selecting the most suitable ones to tackle specific economic and financial scenarios is by no means trivial and calls for a deep awareness of the different language features, their mutual interactions, and how they are implemented by the available systems. Working with the Central Bank of Italy, we have been embarking on this adventure since five years ago and, together with the University of Oxford and TU Wien, have driven the development of Vadalog, one of the state-of-the-art reasoners. In this contribution, we summarize the characteristics of existing reasoners offering a summary table which, to the best of our knowledge, provides an in-depth comparison of all of them for the first time. Leveraging five reference use cases of interest, we discuss the most relevant language features and choices that have driven the design of Vadalog. As a by-product, we offer a guide for other entities in the private and public sectors (e.g., central banks, statistical offices, supervision authorities, financial intelligence units, etc.) willing to enhance their business processes with automated reasoning.