Document Data Matching for Blockchain-Supported Real Estate

The real estate sector remains highly dependent on manual document handling and verification, making processes inefficient and prone to fraud. This work presents a system that integrates optical character recognition (OCR), natural language processing (NLP), and verifiable credentials (VCs) to automate document extraction, verification, and management. The approach standardizes heterogeneous document formats into VCs and applies automated data matching to detect inconsistencies, while the blockchain provides a decentralized trust layer that reinforces transparency and integrity. A prototype was developed that comprises (i) an OCR-NLP extraction pipeline trained on synthetic datasets, (ii) a backend for credential issuance and management, and (iii) a frontend supporting issuer, holder, and verifier interactions. Experimental results show that the models achieve competitive accuracy across multiple document types and that the end-to-end pipeline reduces verification time while preserving reliability. The proposed framework demonstrates the potential to streamline real estate transactions, strengthen stakeholder trust, and enable scalable, secure digital processes.