Automated Label Linking via LSH Forest and ML

The patent replaces the manual mechanical process of linking domain definitions to technical metadata with an automated machine learning system. The system uses trained models to automatically perform the linking that previously required human data stewards to manually search and associate terms, thereby eliminating the time-consuming manual effort while maintaining or improving data governance quality.

The system enables self-service automation where the data catalog automatically performs the linking operation without requiring human intervention. The machine learning models autonomously analyze technical metadata, generate candidate domain definitions, and establish associations, allowing the system to serve itself rather than relying on continuous manual input from data practitioners.

The patent substitutes manual search and linking operations with automated machine learning-based systems. Instead of data practitioners manually searching through growing metadata volumes, the system automatically discovers relationships between technical metadata and domain definitions using trained models, maintaining ease of data discovery regardless of metadata scale.

The system changes the operational parameters from manual human search behavior to automated algorithmic processing. By transforming the data discovery process from human-cognitive operations to machine-executed algorithms, the system achieves scalability that manual processes cannot match, as computational operations can handle large volumes efficiently without proportional increases in time or effort.

The patent applies preliminary action by training machine learning models in advance on curated datasets before deployment. The models are pre-trained to understand the relationships between domain definitions and technical metadata, so that during actual operation, the system can quickly perform linking without requiring complex real-time processing. This upfront preparation reduces the operational complexity of the automated system.

The system introduces trained machine learning models as intermediaries between the raw metadata and the linking operation. Rather than implementing complex rule-based systems or requiring direct human intervention, the patent uses pre-trained models as intermediary components that automatically perform the matching, simplifying the overall system architecture while maintaining high productivity.