Context-Aware Knowledge Graph Subgraph Retrieval for Query Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Knowledge graphs with vast amounts of data face challenges in retrieving relevant information due to numerous irrelevant branch nodes and non-adjacent paths, making it difficult for information retrieval systems to identify key nodes for accurate query responses.
Innovation Solution
Generating context-relationship ranking predictions using predictive machine learning models to extract relevant information from knowledge graphs by identifying candidate node paths and relations, creating subgraph data objects, and generating answer outputs based on these predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If a knowledge graph comprises a large amount of data to support comprehensive information retrieval, then the completeness of information is improved, but the difficulty of identifying key nodes and retrieving relevant data increases
Solution Approach 1:
The patent segments the large knowledge graph into multiple subgraphs based on query context relevance. The system identifies and extracts only the relevant portions of the knowledge graph that pertain to the specific query, dividing the overwhelming full graph into manageable, context-specific segments. This segmentation reduces the search space from the entire knowledge graph to only the necessary subset, making key node identification feasible.
Solution Approach 2:
The patent applies local quality by assigning different processing priorities and attention levels to different regions of the knowledge graph based on their relevance to the query. Contextually relevant nodes and paths receive focused processing and higher priority, while irrelevant regions are either pruned or processed with lower priority. This selective focus allows the system to handle the large data volume by concentrating computational resources on locally relevant areas.
2Adaptability or versatility
If a knowledge graph includes many branch nodes and non-adjacent paths to maintain comprehensive information, then the coverage of information is improved, but the complexity of data retrieval increases
Solution Approach 1:
The patent extracts only the necessary subgraph portions required to answer the specific query, taking out the relevant nodes and edges from the full knowledge graph while leaving the rest untouched. This extraction process removes irrelevant branch nodes and non-adjacent paths that would otherwise complicate the retrieval process, maintaining information coverage for the query at hand while reducing overall retrieval complexity.
Solution Approach 2:
The patent performs preliminary filtering and ranking of candidate node paths based on query context before actual retrieval operations. By pre-identifying and ranking the most relevant paths and nodes before executing the full retrieval process, the system prepares the data structure in advance to minimize the complexity of subsequent retrieval operations, handling only the necessary portions of the complex graph.
3Reliability
If an information retrieval system processes all data in a knowledge graph to ensure comprehensive answers, then the accuracy of answers is improved, but the time and computational resources required increase
Solution Approach 1:
The patent applies partial action by processing only the sufficient subset of knowledge graph data needed to answer the query accurately, rather than exhaustively processing all available data. The system identifies and processes the minimum necessary nodes and paths that provide adequate answer accuracy, avoiding the time waste of processing redundant or overly comprehensive data beyond what is needed for the specific query.
Solution Approach 2:
The patent performs preliminary identification and ranking of relevant nodes and paths before full processing. By pre-filtering and prioritizing candidate elements based on query context, the system prepares the data in advance so that only the most relevant portions require intensive processing, significantly reducing the overall time and computational resources needed while maintaining answer accuracy.
Data Source
AI summary
Various embodiments of the present disclosure provide methods, apparatus, systems, computing devices, computing entities, and/or the like for retrieving a subgraph that is used to generate one or more answer outputs responsive to an input query by: (i) generating one or more context embeddings that are associated with an input query, (ii) identifying one or more candidate node paths and one or more node relations based on a knowledge graph, (iii) identifying, using a predictive machine learning model, one or more context-relationship rankings based on the one or more candidate node paths, the one or more node relations, and the one or more context embeddings, and (iv) generating one or more subgraph data objects based on the one or more context-relationship rankings.