Voice Browsing Dialog Tone Selection Using Confidence Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice browsing systems fail to determine a user's mood or assess the likelihood of achieving a conversation goal, leading to unacceptably low satisfaction and potential abandonment of the goal due to inaccurate mood recognition and response.
Innovation Solution
A confidence scoring system that calculates a score indicative of a conversation's progress towards a user's goal, adjusting dialog acts and tones based on input features and user feedback to enhance engagement and success.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If voice browsing systems present standard dialog acts without mood assessment, then the system operation is simple, but user satisfaction becomes unacceptably low due to inaccurate response to user mood
Solution Approach 1:
The system implements feedback by calculating confidence scores based on user responses and using these scores to dynamically adjust dialog acts and tones. The confidence score feedback loop allows the system to learn from user reactions (e.g., negative responses trigger tone adjustments) and improve subsequent interactions, thereby increasing user satisfaction without requiring complex manual mood assessment mechanisms
Solution Approach 2:
The system performs self-service by automatically calculating confidence scores from conversation data and autonomously selecting appropriate dialog acts and tones based on these scores. This self-adjusting mechanism eliminates the need for external mood assessment tools or complex manual interventions, maintaining system simplicity while improving reliability through automated adaptive behavior
2Reliability
If voice browsing systems use fixed dialog acts without confidence scoring, then the system structure is simple, but the likelihood of achieving conversation goals becomes unacceptably low due to inability to assess conversation progress
Solution Approach 1:
The confidence scoring system provides continuous feedback about conversation progress toward user goals. By calculating confidence scores based on user responses and conversation state, the system can identify when goals are at risk and dynamically adjust dialog acts to improve goal achievement likelihood, creating a self-correcting conversation management mechanism
Solution Approach 2:
The system transitions from static fixed dialog acts to dynamic adaptive dialog selection. Confidence scores enable real-time adjustment of dialog acts and tones based on conversation progression and user responses, allowing the system to adapt its behavior dynamically to maximize goal achievement while maintaining operational simplicity through automated scoring mechanisms
3Reliability
If voice browsing systems do not adjust tones based on user mood, then the conversation flow is fast and simple, but user satisfaction becomes unacceptably low due to mismatched communication style
Solution Approach 1:
The system changes communication parameters (tones and dialog act selections) based on calculated confidence scores that reflect user mood and conversation state. By adjusting tone parameters dynamically - such as becoming more apologetic when confidence scores indicate user frustration - the system maintains fast conversation flow while improving satisfaction through parameter adaptation rather than slowing down the interaction
Data Source
AI summary
Dialog acts (e.g., questions) are selected for voice browsing by a model trained to identify a dialog act that is most likely to lead to a desired outcome. Upon receiving an invocation to begin a conversation, a score indicative of a level of confidence that the conversation will have a successful outcome is determined, and a dialog act is selected based on the score. Subsequently, at each turn of the conversation, the score is updated or a new score is calculated, and a dialog act is selected based on the updated or new score. Confidence scores are calculated based on input features that are determined based on the user who uttered the invocation or responses to dialog acts, as well as a context of the conversation, and provided to a linear model or a machine learning model as inputs.