Schema-Based Governing Label Recommendations Without Data Content Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual assignment of governing labels to large volumes of data is not scalable and prone to errors, and analyzing data content for domain discovery is resource-intensive and often restricted by privacy concerns.
Innovation Solution
Automatically generate governing label recommendations using a schema-level hierarchical path and machine learning techniques, such as multilabel classification, rich textual information, and label co-occurrence, without examining the data content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual assignment of governing labels is used, then labeling accuracy can be maintained through human judgment, but scalability deteriorates and error rates increase for large volumes of data
Solution Approach 1:
The system enables data attributes to be automatically labeled through machine learning models that process schema information independently, without requiring continuous human intervention for each labeling task, thereby achieving both scalability and maintained accuracy
Solution Approach 2:
The patent replaces the mechanical human labeling process with an automated machine learning system that uses schema-based predictions and multilabel classification to generate governing labels, eliminating the scalability limitations of manual assignment while maintaining reliability through algorithmic consistency
2Measurement precision
If data content analysis is performed for domain discovery, then labeling accuracy improves, but computing resource consumption increases and privacy restrictions are triggered
Solution Approach 1:
The system extracts and utilizes schema information as a surrogate for full data content analysis, obtaining sufficient labeling accuracy by processing only the structured schema metadata rather than examining actual data values, thereby reducing computing resource usage and avoiding privacy concerns
Solution Approach 2:
The system performs schema-based predictions and generates governing label recommendations in advance without requiring intensive analysis of actual data content, preparing labeling decisions based on structural information that can be processed with minimal computational resources
Data Source
AI summary
Methods and systems are provided for facilitating generation and/or presentation of governing label recommendations for data. In embodiments described herein, a representation of a data schema associated with a dataset having a plurality of attributes is obtained. A governing label for a particular attribute of the plurality of attributes is identifying, via a machine learning model, based on the representation of the data schema associated with the dataset. Thereafter, a recommendation to assign the governing label to the particular attribute in the dataset is presented.