Unified Metadata Graph for AI Model Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems require significant computational resources and manual effort to access and manage siloed data across disparate locations, leading to inefficiencies, data integrity issues, and system downtimes due to the creation of new data silos and reconfiguration of computing systems.
Innovation Solution
A unified metadata graph system that uses natural language processing and Large Language Models (LLMs) to generate a domain-specific metadata graph, reducing the need for new data silos and reconfiguration by providing a centralized access point for siloed data, optimizing data retrieval times, and preserving data integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If new data silos are created to store copies of data for different computing systems, then data accessibility and system independence are improved, but computational resource usage and system complexity increase
Solution Approach 1:
The patent introduces a metadata graph as an intermediary layer between computing systems and data silos. This metadata graph contains location identifiers that map to data silos and lineage information that tracks data relationships. Instead of creating multiple copies of data across silos, the system uses the metadata graph to efficiently locate and access data from its source, reducing the need for redundant data storage and minimizing system complexity while maintaining data accessibility.
2Speed
If data is copied across multiple data silos for different computing systems, then data retrieval speed is improved, but computational resource usage increases
Solution Approach 1:
The patent implements preliminary action by pre-computing and storing metadata information including location identifiers and data lineage in the metadata graph. This metadata is prepared in advance to enable efficient data location and access. When data retrieval is needed, the system queries the pre-prepared metadata graph to quickly locate data in the appropriate silo, achieving fast retrieval without the need to maintain multiple data copies across silos, thus reducing computational resource usage.
3Reliability
If multiple copies of data are maintained in different data silos, then data availability is improved, but data integrity management becomes more difficult
Solution Approach 1:
The patent implements feedback mechanisms through the metadata graph that tracks data lineage information and location identifiers. The system continuously monitors and updates metadata to reflect the current state of data across silos. This feedback loop enables the system to maintain data availability by knowing where data is located while simultaneously managing data integrity by tracking the single source of truth and relationships between data instances, reducing the complexity of integrity management.
4Productivity
If data silos are reconfigured to consolidate data, then computational resource efficiency is improved, but system downtime and manual effort increase
Solution Approach 1:
The patent implements self-service through automated processes that use the metadata graph to dynamically locate and access data across silos without requiring manual reconfiguration. The system automatically queries the metadata graph to determine the optimal location of data and retrieves it accordingly. This automation eliminates the need for manual data consolidation efforts, preventing system downtime while improving computational resource efficiency by accessing data from its existing locations rather than requiring physical consolidation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system facilitates a process for automatically deploying artificial intelligence (AI) models. The system receives, for a first artificial intelligence (AI) model used by an entity, a first request to deploy the first AI model to make the first AI model available for use in a production environment to process input data and generate corresponding outputs. A first model deployment location for the first model is selected based on a model deployment engine. The system generates scripts to deploy the first AI model to the selected location, then monitors operations parameters associated with the deployment of the first AI model. Based on the values of the operations parameters, the system updates the model deployment engine. In response to a second request to deploy a second AI model, the system uses the updated model deployment engine to select a second model deployment location for the second model.