Natural-Language Computing Action Search with Schema Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computing systems face challenges in efficiently and reliably searching for and performing computing actions due to the large number of possible actions, leading to delays, resource wastage, and inaccuracies, particularly in environments like human capital management systems.
Innovation Solution
A system utilizing a machine learning model with a predefined list of actions and metadata validation to accurately detect user intent, reducing hallucinations and improving reliability by embedding account identifiers into prompts and validating model outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a machine learning model is used to detect user intent, then the system can handle natural language queries more flexibly, but the model may produce hallucinations and inaccurate action recommendations
Solution Approach 1:
The patent introduces an intermediary validation layer between the machine learning model and the final action recommendation. This validation layer checks model outputs against the predefined action schema, ensuring that only valid actions are recommended. The schema acts as a mediator that filters out hallucinated or inaccurate recommendations while preserving the flexibility of natural language processing.
Solution Approach 2:
The system performs preliminary action by defining a complete schema of valid actions before the machine learning model generates recommendations. This predefined schema includes all permissible actions, their parameters, and validation rules. The model's outputs are then checked against this pre-established framework, ensuring reliability while maintaining adaptability.
2Reliability
If the system searches through a large number of possible computing actions, then it can find the desired action, but it introduces delays and consumes excessive computing resources
Solution Approach 1:
The patent segments the large set of computing actions into a structured schema with defined categories, parameters, and relationships. This segmentation allows the system to efficiently navigate and search through actions by leveraging the hierarchical structure and metadata, rather than performing brute-force searches through all possible actions.
Solution Approach 2:
The system changes the parameter representation of actions by using a standardized schema with defined parameters and metadata. This parameterization enables efficient filtering, sorting, and matching of actions based on query requirements, significantly reducing search time and computational resources compared to unstructured action searching.
3Productivity
If the system processes instructions to perform computing actions, then it can execute tasks, but it may select or perform incorrect actions leading to resource wastage
Solution Approach 1:
The patent implements feedback mechanisms where the system validates selected actions against the predefined schema before execution. The validation process provides feedback on whether the selected action and its parameters are correct and appropriate for the given query. This feedback loop prevents incorrect actions from being executed, avoiding resource wastage while maintaining productivity.
Data Source
AI summary
The technical solutions described herein present a computing action search using natural language processing. A system can identify a request containing an executable action associated with a first account identifier of a client system and select a prompt that corresponds to the action, is structured as text including fields, and identifies compatible actions corresponding to the client system or the first account identifier. The system can embed content, including text or metadata, of the first account identifier into the fields of the prompt. The system can provide the prompt to a model and obtain a response from the model indicating a recommended action and a second account identifier associated with the recommended action. The system can validate that the recommended action corresponds to the compatible actions, and the second account identifier corresponds to the first account identifier and execute, responsive to validation, the recommended action for the first account identifier.