Knowledge Graph Enrichment Using Frequency-Targeted AI Prompts

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for enriching knowledge graphs using generative AI are computationally costly and prone to inaccuracies due to irrelevant inferences, leading to unreliable and inconsistent data.

Innovation Solution

A method that generates targeted prompts for a language model based on identified relevant properties, reducing unnecessary inferences by using observation frequency thresholds to focus on specific properties for enrichment, thereby minimizing hallucinations and human intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If generative AI language models are polled to retrieve missing knowledge in a knowledge graph without preconceptions, then the language model can generate properties, but the number of inferences required becomes very costly and the information generated may be inaccurate and inconsistent

Engineering Contradiction:

Improveability to generate missing propertiesVSAvoidaccuracy and consistency of generated information

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system performs preliminary action by identifying and ranking missing properties before polling the language model. The properties are classified by observation frequency and relevance to the entity type, so that only the most relevant properties are queried from the language model, reducing unnecessary inferences and improving reliability

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system changes the parameter of property selection by using observation frequency thresholds to filter which properties are queried. This parameter-based filtering ensures that only properties with sufficient observational support are requested from the language model, reducing hallucinations and improving accuracy

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If generative AI language models are polled without preconceptions concerning desirable properties, then all possible properties can be explored, but computational resource costs and energy consumption increase significantly

Engineering Contradiction:

Improvecompleteness of property explorationVSAvoidcomputational resource costs and energy consumption

Core Design Contradiction:

Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The system applies partial action by querying the language model for only a subset of missing properties rather than all possible properties. Properties are selected based on observation frequency thresholds, so that only the most relevant and frequently observed properties are explored through costly language model inferences

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The system performs preliminary classification and ranking of missing properties by observation frequency before submitting queries to the language model. This preliminary sorting enables the system to focus computational resources on the most promising properties, reducing overall energy consumption

Inventive Principle:

Principle #10Preliminary action

3Reliability

If manual collaborative construction is used to construct the knowledge graph, then the information can be reliable and legitimate, but the work becomes long and tedious particularly when knowledge fields are vast

Engineering Contradiction:

Improvereliability and legitimacy of informationVSAvoidspeed of knowledge graph construction

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system applies self-service by automatically identifying missing properties, ranking them by observation frequency, and querying the language model for their values. This automated enrichment process maintains reliability through systematic property selection while dramatically increasing productivity compared to manual construction

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system uses feedback from the knowledge graph itself by analyzing observation frequencies of properties across existing entities. This internal feedback mechanism guides the automated enrichment process, ensuring that properties with higher observational support are prioritized, maintaining reliability while enabling scalable automation

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20260017444A1Method for communicating with a computer device comprising a knowledge database

Publication Date: 2026.01.15 ORANGE SA

AI summary

A method is described for communicating with a computer device comprising a knowledge database modelling data in the form of a knowledge graph. The method includes, on the device, receiving a first request comprising information relating to an entity of the knowledge graph; commanding the rendering of a web page containing this information; receiving a second request requesting at least one missing property of said entity, from among said rendered information; commanding the rendering of a web page containing a list of missing properties classified by the observation frequency of these properties for other entities of the same type; polling a language model, based on a prompt generated by said device in natural language, with said prompt asking the language model for the value of at least one of the properties in the list; and commanding the rendering of a web page containing that value.