AI Sentiment Threshold Adaptation via Q-Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing artificial intelligence systems face challenges in determining personalized sentiment thresholds for users, as these thresholds can vary significantly between individuals and scenarios, affecting communication efficiency and accuracy.

Innovation Solution

The method involves an AI device that uses a Q-learning approach to dynamically determine sentiment thresholds by analyzing monitoring information and updating Q-values based on user interactions, considering personalized factors, conversation scenarios, and external environments to optimize sentiment recognition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a preset sentiment threshold is used in the artificial intelligence device, then the device can determine user sentiment status, but the sentiment threshold cannot adapt to different users and scenarios

Engineering Contradiction:
Improvesentiment threshold adaptabilityVSAvoidthreshold determination complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements dynamic sentiment thresholds by introducing a Q-learning mechanism that continuously updates thresholds based on user interactions and feedback. The threshold transitions from a static preset value to a dynamic parameter that adapts in real-time according to user sentiment responses, thereby resolving the contradiction between adaptability and complexity.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system employs feedback loops where user sentiment responses are collected, processed, and used to update the Q-value table and subsequent sentiment thresholds. This feedback mechanism enables the system to learn from user interactions and continuously optimize threshold values for different users and scenarios without requiring complex manual configuration.

Inventive Principle:
Principle #23Feedback

2Measurement precision

If the sentiment threshold is optimized for individual users, then communication efficiency and accuracy improve, but the system complexity increases

Engineering Contradiction:
Improvesentiment recognition accuracyVSAvoidpersonalization system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The Q-learning-based system enables self-service personalization where the artificial intelligence device automatically learns and adapts to individual user sentiment patterns through interactions. The system autonomously updates its own threshold parameters based on collected feedback, eliminating the need for complex manual personalization processes while achieving high recognition accuracy.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent changes the sentiment threshold parameter from a fixed preset value to a dynamic variable that is continuously adjusted based on Q-learning outcomes. This parameter transformation allows the system to achieve personalized accuracy for each user while maintaining a relatively simple underlying architecture through standardized learning mechanisms.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3605369B1Method for determining emotional threshold and artificial intelligence device
Publication Date: 2024.08.28 HUAWEI TECH CO LTD
  • EP3605369B1 patent drawingFigure 1
  • EP3605369B1 patent drawingFigure 2~3
  • EP3605369B1 patent drawing

AI summary

This application provides a method for determining a sentiment threshold and an artificial intelligence device. The method includes: step 1: determining, by an artificial intelligence device based on obtained monitoring information, that a sentiment status of a first user is a first sentiment state; step 2: obtaining, by the artificial intelligence device, N actions of a second user; step 3: determining, by the artificial intelligence device, a first action based on a Q-value table; step 4: updating, by the artificial intelligence device, a Q value that corresponds to the first sentiment state and the first action in the Q-value table; and step 5: determining, by the artificial intelligence device, whether an updated Q value is greater than a preset threshold, and if the updated Q value is greater than a preset threshold, determining the sentiment threshold based on the monitoring information, or if an updated Q value is not greater than a preset threshold, repeating step 1 to step 5 until the sentiment threshold is determined. According to the foregoing technical solutions, the sentiment threshold may be optimized by using a Q-learning method, to improve communication efficiency and communication effects in different scenarios.