Plug-In Recipe Execution with ML Data-Type Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing plug-in application recipe management systems require significant user input and do not efficiently generate or execute recipes without manual intervention, limiting their automation capabilities.
Innovation Solution
A system utilizing machine learning and semantic analysis to generate and execute plug-in application recipes (PIARs) with minimal user input, enabling automated background execution and adaptive learning based on data type discovery and mapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If machine learning and semantic analysis are used to generate and execute PIARs automatically, then automation capability and system efficiency are improved, but device complexity increases
Solution Approach 1:
The patent introduces a machine learning model as an intermediary component that automatically maps data types between trigger conditions and actions. This mediator handles the complex task of data type discovery and mapping, reducing the need for manual user input while managing the inherent complexity through a specialized automated component rather than requiring users to understand complex mappings themselves.
Solution Approach 2:
The system implements self-service through automatic data type discovery and mapping capabilities. The machine learning model autonomously analyzes data types from trigger conditions and automatically determines compatible actions without requiring manual user configuration. This self-service approach enables automated background execution of PIARs, significantly improving automation capability while keeping the user interface simple.
2Productivity
If manual user input is required for PIAR generation and execution, then system complexity is reduced, but productivity and automation efficiency deteriorate
Solution Approach 1:
The system enables self-service by automatically discovering data types from incoming trigger conditions and autonomously mapping them to appropriate actions using a machine learning model. This eliminates the need for manual user input in data type mapping, significantly improving automation efficiency. Users simply need to define high-level PIAR logic, while the system handles the complex data type matching automatically.
Solution Approach 2:
The patent implements preliminary action by pre-training the machine learning model with data type mappings and relationships before actual PIAR execution. The system performs data type discovery and compatibility analysis in advance, building a knowledge base that enables rapid automated decision-making during runtime. This preliminary preparation allows the system to execute PIARs efficiently without requiring manual user input during critical execution phases.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques for user-assisted plug-in application recipe (PIAR) execution are disclosed. During execution of a PIAR, a PIAR management application applies one or more data values for a plug-in application field to a machine learning model, to obtain: (a) a candidate mapping between one or more sub-values discovered within the data value(s) and another field accepted by an action of another plug-in application, the data value(s) being of a data type different from a reported data type of the other field, and (b) a confidence metric associated with the candidate mapping, based at least in part on whether the sub-value(s) fit(s) one or more stored formats mapped to the other data type. Based on a determination that the confidence metric does not satisfy a threshold confidence criterion, the PIAR management application obtains user input affirming or rejecting the candidate mapping, and applies the user input to execution of the PIAR.