Generative AI Knowledge Graphs for Reliable Intent Taxonomy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in efficiently and accurately classifying and updating ontologies for digital documents due to their varying categories, leading to incorrect or inefficient query processing.
Innovation Solution
A knowledge graph data structure is constructed using generative artificial intelligence, iteratively generating taxonomies from historical queries and evaluating each level with taxonomy criteria to minimize errors and automate the process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional ontology classification methods are used for digital documents, then the system can maintain a structured classification framework, but it becomes technically challenging to accurately classify documents with increasingly varying categories without introducing excessive errors and computing processing delays
Solution Approach 1:
The patent segments the ontology construction process into hierarchical levels (top-level categories, mid-level categories, leaf categories) that can be independently generated and evaluated. Each level is processed separately through iterative generation and evaluation cycles, allowing the system to handle varying categories at different granularities without overwhelming the classification process.
Solution Approach 2:
The patent implements a dynamic ontology that can be automatically updated and expanded as new document categories emerge. The system continuously iterates through generation and evaluation cycles, allowing the ontology to adapt to increasingly varying categories while maintaining classification accuracy through automated evaluation metrics.
2Reliability
If manual ontology updates are performed to accommodate varying document categories, then the ontology can be maintained, but it becomes challenging to update the ontology in a reliable and efficient manner without introducing excessive errors and delays
Solution Approach 1:
The patent implements self-service ontology updates through automated generation and evaluation. The system automatically generates new categories, evaluates them against evaluation metrics, and updates the ontology without human intervention. This eliminates manual errors and significantly reduces the time required for ontology updates while maintaining reliability through systematic evaluation.
Solution Approach 2:
The patent incorporates feedback mechanisms where evaluation metrics assess generated categories and provide information for iterative improvement. The system uses evaluation results to refine category generation, ensuring reliable ontology updates that adapt to new document categories while minimizing errors and update time.
3Extent of automation
If generative artificial intelligence is used to construct knowledge graph data structures, then the system can automatically generate multi-level taxonomies from historical queries, but it may produce hallucinations or incorrect classifications
Solution Approach 1:
The patent implements feedback loops where generated taxonomies are evaluated against evaluation metrics and historical queries. The system iteratively refines category generation based on evaluation results, correcting hallucinations and improving classification accuracy while maintaining high automation levels.
Solution Approach 2:
The patent performs preliminary evaluation of generated categories before finalizing the taxonomy. By evaluating categories against metrics and historical queries in advance, the system identifies and corrects potential hallucinations before they affect classification accuracy, ensuring reliable automated taxonomy generation.
4Adaptability or versatility
If the knowledge graph structure is expanded to handle increasingly complex query patterns, then the system can adapt to changing configurations, but it becomes challenging to efficiently generate and maintain the knowledge graph tree structure
Solution Approach 1:
The patent segments the knowledge graph into hierarchical levels that can be independently generated and maintained. Each level handles specific aspects of query patterns, allowing the system to adapt to complex queries while maintaining manageable complexity through modular, level-based construction.
Solution Approach 2:
The patent implements a dynamic knowledge graph that automatically adapts to changing query patterns through iterative generation and evaluation. The system expands and contracts the knowledge graph structure based on observed query configurations, maintaining versatility while managing complexity through automated adaptation rather than manual maintenance.
Data Source
AI summary
This application is directed to constructing a knowledge graph using generative artificial intelligence. A system can include one or more processors coupled with memory to identify a plurality of items of unstructured data. The system can provide, for one or more generative artificial intelligence models, a first prompt to cause the models to output a plurality of first level categories of a hierarchical data structure for the items. The system can receive the first level categories, each corresponding to a subset of the items grouped by semantic similarity, and evaluate each category according to taxonomy criteria. The system can provide a second prompt to generate second level categories for each first level category, receive the second level categories, and construct a knowledge graph data structure linking the categories and their respective subsets to relate each item of unstructured data with corresponding categories according to the hierarchical data structure.