Digital Twin Search Using Graph Embeddings for Sparse Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital-twin-oriented search and recommendation systems fail to provide satisfactory numbers, variety, and relevance of search results, particularly when user data is sparse or user queries lack familiarity with the dataset terminology.
Innovation Solution
A computer-implemented method using a trained graph neural network to generate semantic and graph embeddings of digital twins, leveraging their relationships within a graph structure, to identify relevant digital twins based on semantic and structural similarities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional search methods are used to find digital twins, then the search process is simple and fast, but the number, variety, and relevance of search results are insufficient
Solution Approach 1:
The patent introduces graph embeddings as an intermediary representation that bridges the gap between simple keyword matching and comprehensive relevance assessment. By representing digital twins and their relationships in a graph structure and generating embeddings that capture both semantic meaning and relational context, the system mediates between fast traditional search and thorough relevance evaluation, delivering both speed and accuracy.
Solution Approach 2:
The patent transitions from traditional one-dimensional keyword-based search to multi-dimensional search by incorporating graph embeddings that capture semantic relationships, hierarchical structures, and contextual connections. This dimensional expansion allows the system to evaluate digital twins based on multiple criteria simultaneously, significantly improving result relevance while maintaining search efficiency.
2Measurement precision
If collaborative filtering is used to improve search results, then personalized recommendations are provided, but the system fails when user data is sparse
Solution Approach 1:
The patent pre-computes graph embeddings for all digital twins in the dataset, capturing their semantic and relational characteristics before any search query is received. This preliminary action creates a rich representation framework that does not depend on user interaction history, enabling accurate and personalized search results even for new users or sparse query scenarios.
Solution Approach 2:
The system enables digital twins to 'self-describe' their characteristics through graph embeddings that automatically capture their semantic meaning and relationships within the dataset. This self-service mechanism eliminates the need for extensive user profiling or collaborative filtering, as each digital twin's representation is independently optimized based on its inherent properties and connections.
3Speed
If semantic similarity search is used, then results are provided quickly, but the variety and relational relevance of results are limited
Solution Approach 1:
The patent segments the search process into two efficient stages: first, traditional keyword-based filtering to quickly narrow down candidate digital twins, and second, graph embedding-based ranking to precisely rank candidates by relevance. This segmentation maintains search speed while dramatically improving result variety and relational relevance through the embedding comparison stage.
Data Source
AI summary
To search for digital twins (avatars) on a computerized platform, a semantic embedding is made of the search query and this is compared with respective semantic embeddings generated for digital twins on the computerized platform, to determine a first set digital twins whose semantic embeddings are similar to that of the search query. A trained graph neural network generates respective graph embeddings of feature data of the first set of digital twins, and these graph embeddings are compared with graph embeddings that the trained graph neural network generates for other digital twins on the platform, to determine a second set of digital twins whose graph embeddings are similar to graph embeddings of the first set of digital twins. The search results returned in response to the search query may include at least one digital twin in the second set of digital twins.