Adaptive Dialogue System Self-Learning Response Generation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current chatbots and robots have limited capabilities to learn and expand their response repertoire through interaction with human users, leading to answer gaps and inefficiencies in processing user inquiries, relying on fallback reactions and internet searches for responses.

Innovation Solution

A computer-implemented method for an adaptive dialog system that automatically detects and categorizes dialog segments, weighting them by frequency, allowing for the generation and storage of new responses within a dynamic model, reducing the need for external searches and enhancing learning capabilities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If chatbots/robots use conventional knowledge bases with pre-programmed responses, then they can provide automated responses to user queries, but they cannot handle queries that fall outside their programmed scope, leading to answer gaps

Engineering Contradiction:
Improveresponse repertoireVSAvoidknowledge base size
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The dialogue system automatically learns new dialogue segments and generates corresponding responses without human intervention. When encountering unknown dialogue segments, the system autonomously captures them, compares them with the existing model, assigns corresponding segments, and stores them back in the knowledge base, enabling self-expansion of the response repertoire

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system implements a feedback mechanism where dialogue segments that are not found in the knowledge base are captured, analyzed, and used to update the knowledge base. This closed-loop feedback enables continuous improvement of the system's response capabilities based on actual user interactions

Inventive Principle:
Principle #23Feedback

2Reliability

If chatbots/robots use fallback responses to handle unknown queries, then they can maintain conversation flow, but they cannot provide accurate or helpful answers, reducing user satisfaction

Engineering Contradiction:
Improveresponse accuracyVSAvoidconversation continuity
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

Instead of relying on pre-programmed fallback responses, the system autonomously learns from unknown queries by capturing them, processing them through the dialogue segment model, and generating appropriate responses. This self-learning mechanism replaces static fallback responses with dynamically learned accurate responses

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If chatbots/robots perform search engine queries to find answers for unknown dialogue segments, then they can potentially find relevant information, but they require additional hardware resources and time, reducing efficiency

Engineering Contradiction:
Improveanswer coverageVSAvoidprocessing speed
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The system performs preliminary actions by maintaining a structured dialogue segment model that is continuously updated with learned segments. When a query is received, the system first checks this pre-populated model before considering external searches, significantly reducing the need for time-consuming search engine queries

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The dialogue segment model is dynamic and continuously evolves through automatic learning from new dialogue segments. This dynamic adaptation allows the system to improve its answer coverage over time without requiring external search resources, as the knowledge base grows organically from user interactions

Inventive Principle:
Principle #15Dynamics

4Adaptability or versatility

If developers manually program comprehensive response scenarios for chatbots/robots, then they can cover more use cases, but the development time and complexity increase significantly

Engineering Contradiction:
Improveuse case coverageVSAvoiddevelopment effort
Core Design Contradiction:
Adaptability or versatilityVSEase of manufacture

Solution Approach 1:

The system eliminates the need for manual programming of comprehensive response scenarios by automatically learning dialogue segments and generating corresponding responses. The dialogue segment model is self-updating, capturing new segments from user interactions and autonomously integrating them into the knowledge base

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system implements continuous feedback from user interactions to improve its response capabilities. Unknown dialogue segments are captured, processed through the model, and used to update the knowledge base, creating a self-improving system that expands its use case coverage automatically without additional development effort

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3570189B1An adaptive dialogue system and computer-implemented method for providing an adaptive dialogue system
Publication Date: 2023.08.09 SIEMENS AG
  • EP3570189B1 patent drawingFigure 1
  • EP3570189B1 patent drawingFigure 2~3

AI summary

The invention relates to a computer-implemented method for providing an adaptive dialogue system. This involves the automatic detection of at least one dialogue segment from a participant communicating with the dialogue system. The detected dialogue segment is then automatically compared with dialogue segments from a stored dialogue segment model. Following this comparison, at least one corresponding dialogue segment is automatically assigned to the detected segment according to the dialogue segment model, if the detected segment is included in the model. If the detected segment is not included in the model, the detected dialogue segment is automatically added to a group of dialogue segments. A new dialogue segment is then generated based on the other dialogue segments in the group, and this generated segment is stored in the dialogue segment model.