ML-Generated JSON Schema Validation Framework
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
JSON data formats are vulnerable to attacks when used by untrusted services, particularly due to lack of strict validation rules, leading to security threats such as script injection and ineffective validation processes in microservice architectures, which can compromise sensitive data and impact performance.
Innovation Solution
A data validation framework using machine learning to generate schemas dynamically for validating data in microservice environments, providing custom validation injections and security compliance, and allowing for efficient validation processes without repetitive validations across microservices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If JSON data format is used for data exchange between microservices, then data sharing flexibility is improved, but security vulnerability increases due to lack of strict validation rules
Solution Approach 1:
The patent applies preliminary action by generating and storing validation schemas in advance using machine learning models before data exchange occurs. The system pre-processes training data to create validated schemas that are then reused during runtime, eliminating the need for repetitive validation while maintaining security. This resolves the contradiction by preparing validation rules beforehand, allowing flexible JSON exchange without runtime security risks.
Solution Approach 2:
The patent uses copying by creating validated schema copies from training data through machine learning. Instead of validating against original complex data structures repeatedly, the system creates simplified schema copies that capture essential validation rules. These schema copies are then used for efficient validation during data exchange, maintaining security while enabling flexible JSON processing.
2Object-affected harmful factors
If strict validation rules are implemented for JSON data, then security is improved, but validation time and processing cost increase
Solution Approach 1:
The system performs validation schema generation in advance using machine learning models trained on historical data. By pre-processing and storing validated schemas before runtime operations, the system eliminates repetitive validation processing during actual data exchange. This resolves the contradiction by shifting validation work to a preliminary phase, ensuring security without incurring ongoing validation time costs during microservice communications.
Solution Approach 2:
The patent changes the parameter of validation from direct data structure validation to schema-based validation. By transforming complex data validation into simpler schema matching operations, the system maintains strict security requirements while dramatically reducing validation time. The machine learning-generated schemas capture security rules in a format that enables rapid validation during runtime.
3Manufacturing precision
If validation schema is defined for each microservice, then validation accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies universality by creating a centralized machine learning model that generates validation schemas applicable across multiple microservices. Instead of manually defining schemas for each service, a single ML model produces universal schemas that can validate data across different microservice boundaries. This resolves the contradiction by providing high validation accuracy through centralized intelligence while reducing system complexity through schema reusability.
Solution Approach 2:
The system implements self-service by enabling microservices to automatically obtain and use validation schemas generated by the machine learning model without manual configuration. Each microservice can independently validate its data exchanges using schemas that are automatically provided by the validation framework. This resolves the contradiction by maintaining high validation accuracy while reducing operational complexity through automated schema management.
4Productivity
If machine learning is used to generate validation schemas, then validation efficiency is improved, but computational resource consumption increases
Solution Approach 1:
The patent applies preliminary action by performing the computationally intensive machine learning schema generation process in advance, before runtime validation operations. The ML model trains on historical data and generates validation schemas during an off-peak phase, then stores these schemas for rapid reuse during actual data validation. This resolves the contradiction by concentrating computational resource consumption in a preliminary phase, enabling high validation efficiency during runtime without ongoing computational overhead.
Data Source
AI summary
A method, a computer system, and a computer program product for data validation in a microservice environment is provided. Embodiments of the present invention may include receiving a request based on an application configuration validation. Embodiments of the present invention may include determining a schema is not defined based on the request. Embodiments of the present invention may include generating the schema using machine learning. Embodiments of the present invention may include using the generated schema for a plurality of validations.