Fighting crime with Transformers: Empirical analysis of address parsing methods in payment data
This addresses the challenge of location identification in financial transactions for compliance, though it is incremental as it applies existing Transformer methods to noisy real-world data.
The paper tackled the problem of address parsing from free text in payment data to meet regulatory requirements, finding that a fine-tuned Transformer model with early-stopping significantly outperformed other methods, while generative LLMs showed strong zero-shot performance.
In the financial industry, identifying the location of parties involved in payments is a major challenge in the context of various regulatory requirements. For this purpose address parsing entails extracting fields such as street, postal code, or country from free text message attributes. While payment processing platforms are updating their standards with more structured formats such as SWIFT with ISO 20022, address parsing remains essential for a considerable volume of messages. With the emergence of Transformers and Generative Large Language Models (LLM), we explore the performance of state-of-the-art solutions given the constraint of processing a vast amount of daily data. This paper also aims to show the need for training robust models capable of dealing with real-world noisy transactional data. Our results suggest that a well fine-tuned Transformer model using early-stopping significantly outperforms other approaches. Nevertheless, generative LLMs demonstrate strong zero-shot performance and warrant further investigations.