Cognitive Switching Logic for Knowledge Domain Routing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current virtual assistant systems face challenges in efficiently routing user queries across multiple knowledge domains due to the need for complex keyword management and synchronization, leading to suboptimal response accuracy and increased computational costs.

Innovation Solution

A cognitive switching logic is implemented using a neural network that processes natural language information and user feedback to dynamically identify the most relevant knowledge domain, allowing for context-based decision-making and adaptive learning to refine responses, thereby reducing the number of domains queried and improving accuracy over time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If complex keyword management and synchronization are used to route user queries across multiple knowledge domains, then the system can handle diverse queries, but the device complexity and computational costs increase

Engineering Contradiction:

Improvecapability to handle diverse queriesVSAvoidcomplexity of keyword management

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical keyword-based routing system with a neural network-based cognitive switching logic. Instead of manually managing keywords and synchronization mechanisms, the system uses a trained neural network to automatically determine the appropriate knowledge domain based on user queries and context, eliminating the need for complex manual configuration while maintaining the ability to handle diverse queries across multiple domains

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system changes the parameters for query routing from static keyword matching to dynamic neural network inference. The neural network processes input queries along with contextual information and user persona data to produce domain routing decisions, transforming the routing mechanism from rigid keyword-based to flexible learned-based parameter transformation

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If multiple knowledge domains are queried to ensure comprehensive responses, then response accuracy improves, but computational costs and processing time increase

Engineering Contradiction:

Improveresponse accuracyVSAvoidcomputational costs

Core Design Contradiction:

Measurement precisionVSLoss of energy

Solution Approach 1:

The system performs preliminary action by using the neural network to predict and identify the most relevant knowledge domain(s) before actually querying them. The cognitive switching logic processes the user query, retrieves relevant context from previous messages, and proactively determines which knowledge domain(s) are most likely to contain the answer, thereby avoiding unnecessary queries to irrelevant domains and reducing computational waste

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback mechanisms where user responses and interactions are used to refine and retrain the neural network model. This feedback loop allows the system to learn from actual user needs and improve its domain routing accuracy over time, ensuring that fewer and more relevant domains are queried, thus reducing computational costs while maintaining high response accuracy

Inventive Principle:

Principle #23Feedback

3Reliability

If the system queries all knowledge domains to ensure no relevant information is missed, then completeness is improved, but the time required for response generation increases

Engineering Contradiction:

Improvecompleteness of responseVSAvoidresponse generation time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

Instead of querying all knowledge domains (excessive action), the neural network enables the system to perform partial action by identifying and querying only the most relevant knowledge domain(s) based on the user query and contextual information. This selective approach maintains response completeness by focusing on the most likely relevant domains while significantly reducing the time required compared to exhaustive querying of all domains

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS11263527B2Cognitive switching logic for multiple knowledge domains

Publication Date: 2022.03.01 KYNDRYL INC

AI summary

A method provides a set of predicted responses to a user. The method includes receiving a message from a user, the message having natural language information. The method includes processing, using cognitive switching logic (CSL), the natural language information and information from previous messages from the user. The method includes identifying, using CSL, a context of the natural language information based on the information from the previous messages. The method includes identifying, using CSL, at least one knowledge domain which contains a response to the message, based on the identified context and on user persona information. The method includes retrieving a response from each identified knowledge domain. The method further includes, in response to retrieving more than one response, transmitting feedback to CSL to refine identifying the at least one knowledge domain until only one response is retrieved. The method further includes presenting the one response to the user.