Chain-of-Thought Knowledge Graphs for Multi-Session Query Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional AI-based conversational assistants struggle to maintain and update knowledge graphs in real-time, fail to understand and remember conversation context across multiple sessions, and inaccurately predict user queries or topics, leading to a suboptimal user experience.

Innovation Solution

Implementing a chain-of-thought knowledge graph (CoT-KG) that models a user's thought process through a knowledge graph comprising nodes and edges, allowing for personalized responses by constraining machine learning models with a user-specific goal topic, enabling real-time updates and improved context understanding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional knowledge graphs are used to enhance bot understanding, then the bot can better understand context, but the knowledge graphs are difficult to maintain and update in real-time

Engineering Contradiction:

Improvecontext understanding accuracyVSAvoidknowledge graph maintenance complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system enables automatic knowledge graph construction and updating by processing user interaction data through machine learning models. The knowledge graph is self-updating based on user feedback and conversation patterns, eliminating manual maintenance requirements while maintaining high context understanding accuracy

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system incorporates user feedback loops where user interactions, corrections, and engagement patterns continuously feed back into the knowledge graph structure. This feedback mechanism allows the knowledge graph to automatically adapt and update in real-time based on actual user behavior, resolving the maintenance complexity issue while preserving reliability

Inventive Principle:

Principle #23Feedback

2Productivity

If machine learning models are used to predict user queries, then the bot can provide proactive responses, but the prediction accuracy remains insufficient

Engineering Contradiction:

Improveresponse efficiencyVSAvoidquery prediction accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary analysis of user behavior patterns, conversation history, and contextual data to pre-compute likely user queries and topics. By preparing prediction models in advance based on accumulated user data, the system can provide accurate proactive responses without compromising prediction precision

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system replaces simple rule-based prediction mechanisms with advanced machine learning models that analyze patterns in user behavior, conversation context, and interaction history. This substitution enables significantly improved prediction accuracy while maintaining high response efficiency through automated model inference

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If the bot is designed to remember context across multiple sessions, then user experience improves, but system complexity and data processing requirements increase

Engineering Contradiction:

Improveconversation context retentionVSAvoidsystem architecture complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system extracts only the essential contextual elements and user preferences from multi-session conversations, storing them as condensed user profiles and context vectors. By extracting only the most relevant information rather than storing complete conversation histories, the system maintains context retention capability while reducing architectural complexity

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system segments user context into distinct modular components such as user preferences, conversation topics, interaction patterns, and contextual entities. This segmentation allows the system to manage and retrieve specific context elements independently, reducing overall system complexity while maintaining comprehensive context awareness across sessions

Inventive Principle:

Principle #1Segmentation

Data Source

PatentEP4700603A1User profiling using chain-of-thought knowledge graphs for querying a machine learning system

Publication Date: 2026.02.25 EQUINIX INC

Figure 1
Figure 2
Figure 3

AI summary

Techniques are disclosed for a machine learning model, such as a large learning model (LLM), that incorporates a model of a chain of thought of a particular user when responding to a query from the user. In one example, a system generates a knowledge graph of a chain of thought of the user. The knowledge graph comprises nodes representing topics present within past queries by the user and edges representing a co-occurrence between the topics. The system determines, based on a topic present within a query from the user and the knowledge graph, a goal query comprising a goal topic. The system provides, to a machine learning model, the user to generate, by the machine learning model, a response. The machine learning model is constrained to include the goal topic of the goal query within the response. The system outputs, for display, the response to the query.