Automated Process Model Validation via Staged NLP Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing process modeling systems face challenges in efficiently validating model notations, as it can be computationally expensive, impossible to validate entire models, and often results in false negatives.
Innovation Solution
An automated process modeling system that uses natural language processing (NLP) to generate model notations in formats like BPMN, CMMN, and DMN, and includes a method for comparing envisioned and deployed processes to highlight differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional validation methods are used to validate model notations, then validation thoroughness is improved, but computational cost increases significantly
Solution Approach 1:
The patent segments the validation process into multiple stages: syntactic validation, semantic validation, and execution validation. Each stage processes different aspects of the model notation independently, allowing thorough validation while managing computational resources efficiently through staged processing rather than exhaustive single-pass validation
Solution Approach 2:
The patent implements selective validation that performs comprehensive checks on critical path elements and essential model components, while applying lighter validation to less critical areas. This partial action approach ensures sufficient validation thoroughness for safety-critical elements without incurring the computational cost of exhaustive validation of all model aspects
2Reliability
If comprehensive validation of entire model notations is attempted, then validation completeness is improved, but validation time increases excessively
Solution Approach 1:
The patent performs preliminary syntactic validation and structural checks before executing time-consuming semantic and behavioral validations. By pre-processing and filtering models to identify only those requiring intensive validation, the system achieves comprehensive validation of necessary elements while significantly reducing overall validation time through early elimination of obviously valid or invalid models
Solution Approach 2:
The patent applies different validation depths to different parts of the model notation based on their criticality. Critical path elements, decision logic, and safety-relevant components receive exhaustive validation, while peripheral or non-critical elements receive streamlined validation, achieving completeness where needed without the time penalty of uniform comprehensive validation
3Measurement precision
If traditional testing approaches are used for model validation, then detection capability is improved, but false negative rate increases
Solution Approach 1:
The patent implements feedback mechanisms where validation results from syntactic, semantic, and execution stages are continuously fed back to refine subsequent validation efforts. Detection patterns from initial testing inform adaptive adjustment of validation criteria, allowing the system to learn from false negatives and improve detection accuracy iteratively, thereby reducing false negative rates while maintaining high detection capability
Solution Approach 2:
The patent employs dynamic validation thresholds and detection criteria that adapt based on model characteristics, historical validation data, and risk profiles. Rather than static testing approaches, the system dynamically adjusts its detection sensitivity and validation depth, improving false negative detection in high-risk areas while avoiding over-validation in low-risk areas, thus maintaining high detection capability with reduced false negatives
Data Source
AI summary
Disclosed herein are system, method, and computer program product embodiments for machine-assisted process modeling and validation. An embodiment operates by receiving, by at least one processor, a process document describing a process in a user locale. The embodiment then generates the model notation in accordance with a model notation format by processing the process document with a deep learning technique based on a prompt for modeling the process document. The embodiment then outputs the model notation.