Neural Network Data Purging for Enterprise Workflow Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data purging techniques in enterprise systems are inflexible and inefficient, as they rely on static temporal schedules and fail to discriminate among different types of job-related data, leading to system burdens and performance issues.
Innovation Solution
A deep neural network is trained on multiple types of attribute data from historical workflow data to dynamically analyze workflows, determine relevant data for purging, and automatically remove unnecessary data structures from databases based on the neural network's output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional static temporal schedule purging is used, then implementation simplicity is maintained, but data purging effectiveness and system performance improve
Solution Approach 1:
The patent replaces the mechanical/temporal schedule-based purging system with an intelligent system using machine learning models. The ML models analyze workflow attributes and job-related data to dynamically determine purging decisions, substituting rigid temporal rules with adaptive intelligent analysis that improves purging effectiveness while maintaining reasonable system complexity.
Solution Approach 2:
The system changes the parameters used for purging decisions from simple temporal schedules to multiple workflow attributes and job-related data characteristics. By analyzing diverse parameters such as workflow type, job status, and data importance, the system achieves more effective and nuanced data purging that adapts to actual system conditions.
2Adaptability or versatility
If static temporal schedule purging is used, then implementation simplicity is maintained, but discrimination among job-related data types is achieved
Solution Approach 1:
The patent applies local quality by treating different types of job-related data differently based on their specific characteristics. Instead of applying a uniform temporal schedule to all data, the ML models analyze individual data attributes and apply differentiated purging decisions tailored to each data type's importance, sensitivity, and workflow context.
Solution Approach 2:
The system transitions from a single temporal parameter to multiple attributes and characteristics for data evaluation. By considering diverse parameters such as data type, workflow stage, and business importance, the system achieves fine-grained discrimination among job-related data types with adaptable purging strategies.
3Reliability
If more job-related data is retained, then data availability is improved, but system performance and operational efficiency deteriorate
Solution Approach 1:
The patent applies partial action by selectively purging only the portion of job-related data that is determined to be unnecessary through ML analysis. Rather than implementing blanket retention or blanket deletion policies, the system performs targeted purging that removes only the excess data while preserving essential information, achieving the optimal balance between availability and efficiency.
Data Source
AI summary
Methods, apparatus, and processor-readable storage media for automatically purging data using a deep neural network are provided herein. An example computer-implemented method includes training a neural network model using multiple types of attribute data and job-related data associated with historical workflow data maintained within multiple data structures of an enterprise system; dynamically analyzing workflows derived from the enterprise system, wherein dynamically analyzing the workflows comprises determining the multiple types of attribute data for the workflows and processing job-related data associated with the workflows into multiple data structures within one or more databases; applying the neural network model to the determined attribute data; and removing at least a portion of the multiple data structures from the one or more databases based at least in part on the application of the neural network model to the determined attribute data.