Conversation Sentence Generation Device for Human-like Dialogue
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional dialogue systems struggle to generate human-like conversations, as they are limited to predetermined tasks and scenarios, resulting in uniform and unvaried dialogue that fails to account for the physical and psychological states of conversation partners.
Innovation Solution
A conversation-sentence generation device that estimates the physical and psychological states of both the agent and user, generating conversation sentences based on these states and utterance intentions, allowing for varied and contextually appropriate responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional dialogue systems use predetermined scenarios and task-based dialogue control, then the system can achieve specific functional goals, but the conversation becomes uniform and lacks human-like variety
Solution Approach 1:
The patent segments the dialogue generation process into distinct modules: intention recognition unit, state estimation unit, and dialogue generation unit. This segmentation allows each module to specialize in specific functions while maintaining overall system coherence, enabling both reliable task completion and varied conversation through independent optimization of each component.
Solution Approach 2:
The patent introduces dynamic state estimation that adapts to changing conversation contexts. The state estimation unit continuously updates the agent's physical and psychological states based on ongoing dialogue, allowing the system to dynamically adjust its response style and content to maintain both task reliability and conversational variety.
2Productivity
If dialogue systems focus on task-oriented conversations, then efficiency is improved, but the system cannot generate diverse utterances suited for different physical and psychological states
Solution Approach 1:
The patent applies local quality by estimating specific physical and psychological states (e.g., hunger, tiredness, happiness) and generating utterances tailored to each state. Each state has its own characteristic expression patterns, allowing the system to maintain high productivity for task completion while adapting utterance style to match the agent's current state.
Solution Approach 2:
The patent changes the parameter space of dialogue generation by incorporating physical and psychological state parameters. Instead of generating utterances based solely on task requirements, the system adjusts utterance characteristics based on state parameters, enabling diverse yet contextually appropriate responses without sacrificing dialogue efficiency.
3Reliability
If the system generates consistent responses based on fixed dialogue scenarios, then task completion is reliable, but previous conversation context and remembered states are not utilized
Solution Approach 1:
The patent implements feedback mechanisms where the state estimation unit continuously monitors and updates the agent's states based on conversation history. This feedback loop ensures that previous context and remembered states are utilized in generating current responses, maintaining both consistency with past remarks and adaptability to evolving conversation contexts.
Solution Approach 2:
The system performs preliminary state estimation and context analysis before generating dialogue responses. By pre-processing and storing relevant context information and state estimates, the system can quickly retrieve and utilize this information during conversation, ensuring both consistency and context awareness without compromising response efficiency.
Data Source
AI summary
A conversation-sentence generation device according to the invention of this application includes: an input unit that receives, as input information, a conversation sentence given from a user to an agent, and clue information based on which a physical and psychological state of the agent is estimated; an agent state storing unit that stores the physical and psychological state of the agent as an agent state; an agent state estimating unit that estimates a new agent state based on the input information and the agent state; an utterance intention generating unit that generates, based on the input information and the agent state, an utterance intention directed from the agent to the user; a conversation sentence generating unit that generates, based on the input information, the agent state, and the utterance intention, a conversation sentence given from the agent to the user; and an output unit that outputs the conversation sentence generated by the conversation sentence generating unit.