Language Model Response Handling for Overlapping User Utterances

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing language models in information processing systems suffer from reduced responsiveness during interactions, requiring users to wait for responses to previous inputs before making new queries.

Innovation Solution

An information processing apparatus and method that allows for the generation and presentation of responses to subsequent user inputs while previous responses are being processed, utilizing a language model trained by machine learning to facilitate immediate follow-up interactions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If the system waits for completion of a first response before processing a second utterance, then the response accuracy is maintained, but the interaction responsiveness deteriorates

Engineering Contradiction:

Improvewaiting timeVSAvoidresponse accuracy

Core Design Contradiction:

Loss of timeVSReliability

Solution Approach 1:

The system starts generating the second response in advance while the first response is still being presented, rather than waiting for completion. This preliminary action reduces the waiting time for the user while maintaining response accuracy through proper interruption handling.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system dynamically adjusts its behavior based on real-time conditions - it can interrupt the first response generation when a second utterance is detected, and dynamically manage multiple response generations concurrently to optimize both responsiveness and accuracy.

Inventive Principle:

Principle #15Dynamics

2Productivity

If the system processes only one utterance at a time sequentially, then the response accuracy is ensured, but the interaction speed deteriorates

Engineering Contradiction:

Improveinteraction speedVSAvoidresponse accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system maintains continuous useful action by overlapping the processing of multiple utterances. While the first response is being presented, the system continues to process the second utterance and generates the second response concurrently, eliminating idle waiting time and improving interaction speed without sacrificing accuracy.

Inventive Principle:

Principle #20Continuity of useful action

Solution Approach 2:

The system performs preliminary processing of the second utterance while the first response is still underway, preparing the second response in advance. This allows the system to maintain high productivity by keeping all processing units continuously busy.

Inventive Principle:

Principle #10Preliminary action

3Loss of time

If the system allows interruption during response generation, then the responsiveness is improved, but the system complexity increases

Engineering Contradiction:

Improveresponse timeVSAvoidsystem complexity

Core Design Contradiction:

Loss of timeVSDevice complexity

Solution Approach 1:

The system segments the response generation process into manageable units, allowing interruption at specific points. By dividing the generation process and managing multiple responses independently, the system reduces the complexity of handling interruptions while still enabling responsive interaction.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250348689A1Information processing apparatus, response method, and storage medium

Publication Date: 2025.11.13 NEC CORP

AI summary

In order to responsiveness in an interaction using a language model, in an information processing apparatus, in a case where a reception unit receives a second utterance in a period from a time when a first utterance is received to a time when presentation of a first response to the first utterance is completed, a generation control unit starts generation of a second utterance based on the second utterance and a presentation unit presents, to a user, the second response generated.