Probabilistic Graphical Model for Knowledge Graph Path Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing knowledge graphs face challenges in identifying important paths and areas of interest, as users find it difficult to navigate and extract insights from the vast number of nodes and edges.
Innovation Solution
The method involves training a probabilistic graphical model using a knowledge graph and a node-path matrix, which is generated through path-based simulations. This model enables probabilistic reasoning to discover important paths and areas of interest within the knowledge graph.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If users directly navigate and extract insights from the knowledge graph, then they can access the raw data, but they find it difficult to identify important paths and areas of interest due to the vast number of nodes and edges
Solution Approach 1:
The patent introduces a probabilistic graphical model as an intermediary between the knowledge graph and the user. This model is trained using path-based simulations that randomly traverse the knowledge graph and record node co-occurrences. The trained model then serves as a mediator that can efficiently answer path-related queries and identify important paths without requiring users to manually navigate the entire knowledge graph, thus resolving the contradiction between ease of navigation and difficulty of detecting important paths
Solution Approach 2:
The patent performs preliminary path-based simulations during the training phase to pre-compute and store the probabilistic relationships between nodes. By conducting these simulations in advance and capturing node co-occurrence patterns in the trained probabilistic model, the system prepares the necessary information beforehand, making it easily accessible during actual queries without requiring users to perform complex navigation tasks
2Quantity of substance
If the knowledge graph contains a vast number of nodes and edges to provide comprehensive data, then it provides extensive information, but users find it difficult to extract insights
Solution Approach 1:
The patent extracts the essential path-related information from the vast knowledge graph by performing random walks and recording node co-occurrences during training. The probabilistic graphical model captures only the relevant relational patterns needed for path queries, effectively extracting useful information while filtering out unnecessary complexity. This allows the system to maintain comprehensive data in the knowledge graph while providing streamlined insight extraction through the trained model
3Reliability
If path-based simulations are run to generate the node-path matrix for training, then the model can learn the underlying distribution, but computational resources and time are required
Solution Approach 1:
The patent employs random walks as a partial sampling method rather than exhaustively analyzing all possible paths in the knowledge graph. By performing a sufficient number of random walks to capture the underlying distribution patterns without enumerating every possible path, the system achieves reliable probabilistic reasoning while significantly reducing training time and computational resources compared to complete path enumeration
Data Source
AI summary
The present disclosure relates to methods and systems that perform probabilistic reasoning on knowledge graphs. The systems and methods use path-based simulations over a knowledge graph to convert the knowledge graph into a probabilistic graphical model that supports probabilistic reasoning on the knowledge graph. The systems and methods use the probabilistic graphical model to discover paths of the knowledge graph in response to a query.