PayNER: A Named Entity Recognition Framework for Financial Payment Data Using Transformer-Based NLP

Named Entity Recognition (NER) has emerged as a critical component in automatingnancial transaction processing, particularly in extracting structured information from un-structured payment data. This paper presents a comprehensive analysis of state-of-the-artNER algorithms speci?cally designed for payment data extraction, including ConditionalRandom Fields (CRF), Bidirectional Long Short-Term Memory with CRF (BiLSTM-CRF),and transformer-based models such as BERT and FinBERT. We conduct extensive exper-iments on a dataset of 50,000 annotated payment transactions across multiple paymentformats including SWIFT MT103, ISO 20022, and domestic payment systems. Our ex-perimental results demonstrate that ?ne-tuned BERT models achieve an F1-score of 94.2%for entity extraction, outperforming traditional CRF-based approaches by 12.8 percentagepoints. Furthermore, we introduce PaymentBERT, a novel hybrid architecture combiningdomain-speci?c ?nancial embeddings with contextual representations, achieving state-of-the-art performance with 95.7% F1-score while maintaining real-time processing capabilities. Weprovide detailed analysis of cross-format generalization, ablation studies, and deploymentconsiderations. This research provides practical insights for ?nancial institutions implement-ing automated sanctions screening, anti-money laundering (AML) compliance, and paymentprocessing systems.