AI Rule Embedding for Electronic Submission Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software applications face challenges in seamlessly processing electronic datasets due to errors, omissions, and incorrectly formatted data, requiring extensive manual coding and resources, especially for large and frequently modified rule sets.
Innovation Solution
A method involving embedding representations of rules generated by a machine learning model, which automatically identifies and evaluates electronic datasets against relevant rules, reducing the need for manual code updates and enabling real-time system updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manually-written software code is used to define acceptable ranges and formats for submissions, then data processing reliability is improved, but development time and resources increase significantly
Solution Approach 1:
The patent uses large language models to generate software code that defines acceptable ranges and formats for data submissions. Instead of manually writing code, the system copies/adapts code patterns from training data and natural language specifications, significantly reducing development time while maintaining reliability through the model's learned understanding of data validation requirements
Solution Approach 2:
The system enables self-service code generation where the large language model automatically creates validation code based on natural language inputs and historical data patterns. The model serves itself by learning from extensive training data and autonomously generating appropriate code without requiring manual programming expertise, thus reducing both development time and resource requirements
2Measurement precision
If manually-written software code is used to ensure compliance with rule sets, then data validation accuracy is improved, but maintenance complexity increases for large and frequently modified rules
Solution Approach 1:
The patent implements a dynamic code generation system where the large language model continuously adapts code based on updated natural language specifications and changing data requirements. When rules are modified, the system dynamically regenerates the appropriate validation code without requiring manual updates to complex codebases, thus maintaining high validation accuracy while reducing maintenance complexity
Solution Approach 2:
The system copies and adapts code patterns from its training data and existing successful implementations when generating validation code. This allows the system to maintain high validation accuracy by leveraging proven code patterns while simplifying maintenance, as changes can be made by updating natural language specifications rather than modifying complex existing code structures
3Reliability
If extensive manual coding is performed to prevent errors in submissions, then submission quality is improved, but computing resources are consumed excessively
Solution Approach 1:
The patent performs preliminary code generation using large language models trained on extensive datasets of validation rules and patterns. By pre-training the model on comprehensive data and having it generate code before actual submission processing, the system achieves high submission quality while reducing the computing resources needed during runtime, as the heavy lifting of code creation occurs during the initial generation phase
Solution Approach 2:
The system copies efficient code patterns from its training data and generates optimized validation code that achieves high submission quality with reduced computational overhead. The large language model learns resource-efficient coding patterns during training and applies them when generating validation logic, thus maintaining quality while consuming fewer computing resources
Data Source
AI summary
Aspects of the present disclosure relate to automated evaluation of electronic datasets. Embodiments include receiving one or more rules related to evaluation of electronic datasets. Embodiments further include generating, via an embedding model, embedding representations of the one or more rules. Embodiments further include receiving an electronic dataset. Embodiments further include identifying a rule that is applicable to the electronic dataset based on using a machine learning model configured to search the embedding representations of the one or more rules based on the electronic dataset. Embodiments further include evaluating, using the machine learning model or an additional machine learning model, the electronic dataset based on the identified rule. Embodiments further include using the machine learning model or the additional machine learning model to generate an evaluation summary for the electronic dataset based on determining that an item within the electronic dataset does not comply with the identified rule.