Natural Language Input Processing Disambiguation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
When a user provides a natural language input that corresponds to multiple enabled skill systems, there is a difficulty in determining which skill system should be invoked to execute the intent, as multiple systems are capable of executing the same intent, leading to ambiguity.
Innovation Solution
A system is configured to determine the confidence scores of multiple skill systems capable of executing a natural language input, and if the difference in scores is less than a threshold, it prompts the user to disambiguate which skill system to invoke, allowing user confirmation or selection, and once a sufficient score difference is achieved, it confidently determines which skill system to execute the intent.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple skill systems are enabled to execute the same intent, then the system's adaptability and versatility are improved, but ambiguity arises in determining which skill system should be invoked
Solution Approach 1:
The system implements a feedback mechanism by evaluating confidence scores from multiple skill systems and using this feedback to determine whether to invoke a skill system directly or to request user disambiguation. When confidence scores are similar (within a threshold), the system seeks additional feedback through user input to resolve ambiguity, thereby maintaining versatility while preventing information loss.
Solution Approach 2:
The patent introduces an intermediary decision-making layer that sits between the natural language input and the skill system invocation. This intermediary evaluates confidence scores from multiple skill systems and mediates the selection process by either directly invoking the highest-confidence system or presenting disambiguation options to the user, thus resolving the conflict between versatility and ambiguity.
2Speed
If the system automatically selects a skill system based on confidence scores, then processing speed is improved, but user control and precision are reduced
Solution Approach 1:
The system applies partial automation by automatically selecting skill systems only when confidence scores show a clear winner (exceeding a threshold margin). When confidence scores are close, the system partially automates by presenting limited options to the user rather than fully automating the selection, thus balancing speed with precision.
Solution Approach 2:
The patent dynamically changes the decision parameter from a fixed automatic selection rule to a variable threshold-based approach. By adjusting the confidence threshold parameter, the system can adapt between automatic selection (higher speed) and user confirmation (higher precision) based on the specific situation, resolving the contradiction between speed and accuracy.
3Measurement precision
If the system requests user disambiguation when confidence scores are similar, then selection accuracy is improved, but processing time increases
Solution Approach 1:
The system applies local quality by requesting user disambiguation only in specific local cases where confidence scores are similar (within a threshold), rather than universally for all intents. This localized application of user interaction minimizes time loss while maintaining high accuracy where it matters most, resolving the contradiction between precision and time efficiency.
Data Source
AI summary
Techniques for intelligently selecting a component to execute with respect to a natural language user input are described. A natural language processing (NLP) system may receive first data representing a natural language input. The NLP system may determine first and second scores representing first and second confidences that first and second components are to be invoked to perform actions responsive to the natural language input, respectively. Based on the first and second scores, the NLP system may determine further information is needed to determine which of the first or second component is to be invoked. The NLP system may query a user for the further information. Based on the further information, the NLP system may determine third and fourth scores representing third and fourth confidences that the first and second components are to be invoked to perform actions responsive to the natural language input, respectively. The NLP system may determine the third score is greater than the fourth score and, based thereon, cause the first component to perform an action responsive to the original natural language input.