Quick Response Estimation Model for Natural Conversation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional conversation systems struggle to generate natural quick responses as a listener, limiting their ability to respond appropriately and naturally in conversations.
Innovation Solution
A learning device and estimation device that acquire speech data and information about speakers and listeners, and use this data to create a learned model that estimates the type of quick response a listener should provide, based on the conversation and classification labels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a conversation system uses random generation for quick responses, then the system can provide quick responses, but the naturalness and appropriateness of the responses deteriorates
Solution Approach 1:
The system pre-collects and stores multiple types of quick response patterns (e.g., acknowledgment responses, agreement responses, transition responses) in a database during the preliminary learning phase. When a quick response is needed, the system retrieves and selects from these pre-prepared patterns based on the conversation context, rather than generating randomly. This resolves the contradiction by having responses ready in advance (maintaining speed) while ensuring they are contextually appropriate (improving naturalness).
Solution Approach 2:
The system changes the parameter of response selection from random selection to context-based selection using a learned model. The learned model analyzes conversation context parameters (speaker information, listener information, conversation flow) and dynamically selects the most appropriate quick response type. This transformation maintains the speed advantage of quick responses while dramatically improving their naturalness and appropriateness through intelligent parameter-based selection.
2Reliability
If a conversation system uses a learned model to estimate quick response types, then the naturalness of responses improves, but the system complexity increases
Solution Approach 1:
The system segments the quick response generation process into distinct functional modules: (1) a learned model for estimating quick response types based on conversation context, (2) a database storing multiple types of quick response patterns, and (3) a selection module that retrieves appropriate patterns. This segmentation allows each module to perform its specific function efficiently, reducing overall system complexity while maintaining high naturalness through specialized processing in each segment.
Solution Approach 2:
The system introduces an intermediary learned model that acts as a bridge between the conversation context and the quick response selection. This learned model estimates the appropriate quick response type by analyzing conversation parameters and mediates between the raw conversation data and the response database. This intermediary layer simplifies the overall system architecture by providing a clear interface between context analysis and response retrieval, reducing complexity while improving response naturalness.
Data Source
AI summary
A learning device includes processing circuitry configured to acquire speech data of a speaker, information on the speaker, conversation data of a listener, information on the listener, and a classification label of a quick response included in the conversation data of the listener, and create a learned model of estimating a type of the quick response of the listener to a conversation of the speaker using the information acquired with the classification label of the quick response as correct answer data.