Operational Data Graphs for Structured-Data Language-Model Insights

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing large language models are not optimized for analyzing structured or semi-structured operational data, making it challenging to extract insights efficiently and accurately from large volumes of operational data generated by systems like sensors and enterprise resource planning systems.

Innovation Solution

Transform semi-structured operational data into textual descriptions using a large language model, apply a data model to generate insights, and create visualizations based on user queries, eliminating the need for manual human analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual human analysis is used to extract insights from operational data, then accuracy can be maintained through human judgment, but time consumption increases significantly and human error is introduced

Engineering Contradiction:

Improveinsight extraction accuracyVSAvoiddata analysis time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces the mechanical human analysis process with an automated computational system comprising data preprocessing modules, graph neural network models, and insight generation algorithms. This substitution eliminates human error and time consumption while maintaining or improving accuracy through consistent automated processing of operational data.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces intermediate processing layers including data preprocessing modules that clean and standardize operational data, graph neural networks that transform data into meaningful representations, and insight generation algorithms that bridge raw data to actionable insights. These intermediaries enable accurate automated analysis without requiring direct human intervention.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Productivity

If large language models are used to analyze operational data, then processing speed and automation improve, but the models are not optimized for structured or semi-structured data sets

Engineering Contradiction:

Improvedata processing speedVSAvoidmodel performance on structured data

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent applies specialized processing techniques tailored to the specific characteristics of operational data. Graph neural networks are used to capture relational structures in the data, while preprocessing modules address the semi-structured nature of operational data. This localized optimization ensures high reliability for structured data analysis while maintaining fast processing speeds.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent transforms operational data into different parameter representations suitable for automated analysis. Data is converted into graph structures with nodes and edges representing entities and relationships, and further transformed into vector representations that preserve structural information while enabling efficient computational processing by automated models.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If automated systems are implemented to eliminate manual data parsing, then efficiency and cost savings are achieved, but system complexity increases

Engineering Contradiction:

Improveanalysis efficiencyVSAvoidsystem architecture complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent divides the automated analysis system into distinct functional modules: data preprocessing modules for cleaning and standardizing input data, graph neural network models for transforming data into relational representations, and insight generation algorithms for producing actionable outputs. This segmentation manages complexity by making each component independent and well-defined.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent designs a multi-functional automated system that can handle various types of operational data (sensor data, transaction data, log data) through a unified architecture. The graph neural network framework provides universal applicability across different data sources and analysis tasks, reducing overall system complexity despite the diverse processing requirements.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250328830A1Systems and methods for generating insights into operational data using a language model

Publication Date: 2025.10.23 PWC PRODUCT SALES LLC

AI summary

A method for generating insights into operational data using a language model includes receiving a user query; receiving summarized operational data, wherein summarized operational data is generated by: receiving operational data; generating an operational data graph, wherein the operational data graph comprises a plurality of nodes and edges; generating a plurality of vectors describing relationships between the plurality of nodes and edges; and applying a data model to the plurality of vectors to generate a natural language description of the plurality of vectors; generating, based on the user query and the summarized operational data, a prompt for querying a first large language model; transmitting the prompt to the first large language model; receiving a natural language response to the prompt; and generating, based on the natural language response and one or more properties of the operational data graph, one or more visualizations corresponding to the natural language response.