CR-ONTO: Ontology-Based Reasoning for Credit Default Classification

Context: Credit defaults are operations less explored by financial institutions compared to performing and non-performing loans. This is because recovering credit default is often ineffective and involves high collection costs. Problem: The literature presents studies that describe ontologies to represent knowledge about loans, but they do not address the conceptualization of default. Furthermore, the ontologies do not explore the possibility of recovering these credits in default condition. Solution: This article presents CR-Onto, an ontology designed for the classification of loans in general, with an emphasis on default. CR-Onto can represent the knowledge of the context involving the credit default and respond to questions suggesting whether a credit default can be settled. SI Theory: CR-Onto is the result of an interdisciplinary study that integrates concepts from finance, computing, and knowledge representation. By organizing data into an ontology, performing inferences, and executing queries, the data is transformed into information that can help credit advisors at a financial institution make better decisions. Method: The ontology has axioms and semantic rules used to provide queries and inferences about its instantiated base. The results were analyzed using a quantitative approach. Summary of results: A synthetic dataset based on a financial institution was constructed to simulate the lifecycle of 5 banking clients. This dataset populated 35 distinct instances of the CR-Onto ontology. This enabled the execution of DL Query to answer 6 competency questions. Contributions and Impact in the IS area: CR-Onto is capable of identifying which credit operations have a higher probability of being resolved. A specialist can utilize these classification results as part of an Information System within a decision support platform.