Temporal Hypergraph Storage for Event Query and Reasoning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing knowledge repositories lack flexible and effective methods for storing and querying event data with temporal characteristics, particularly in knowledge graphs, due to limited temporal analysis and event reasoning capabilities.
Innovation Solution
A hypergraph data storage and query method where events corresponding to multiple nodes are converted into hyperedges, stored in a structured format, and managed within a file system with time windows, allowing for efficient temporal access and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing knowledge repositories are used to store event data, then data storage is simple, but temporal analysis and event reasoning capabilities are limited
Solution Approach 1:
The patent segments event data into discrete temporal units using time windows (e.g., hourly, daily, monthly partitions). Each time window contains events occurring within that specific period, allowing temporal queries to efficiently access only relevant time ranges without scanning entire datasets. This segmentation enables sophisticated temporal analysis while maintaining manageable storage complexity through organized time-based partitions.
Solution Approach 2:
The patent introduces a temporal dimension to traditional knowledge graph storage by organizing events along time axes. Events are stored with temporal metadata including occurrence time, time windows, and sequence numbers, transforming flat storage into multi-dimensional temporal storage. This enables temporal reasoning capabilities while preserving the underlying graph structure through hypergraph representations that capture evolving relationships over time.
2Adaptability or versatility
If hypergraph data with temporal characteristics is stored in a structured format, then temporal analysis capability is improved, but storage and query complexity increases
Solution Approach 1:
The patent segments hypergraph data into time-window-based partitions, where each partition contains events and their relationships within a specific temporal range. This segmentation allows the system to handle complex temporal reasoning by processing smaller, time-bound subsets rather than the entire hypergraph at once, reducing computational complexity while maintaining rich temporal analysis capabilities.
Solution Approach 2:
The patent performs preliminary organization of hypergraph data during the storage phase by pre-computing and storing temporal metadata such as time windows, event sequences, and relationship timestamps. This preliminary structuring enables efficient temporal queries without requiring complex runtime computations, as the temporal framework is already established when data is ingested.
3Loss of information
If events are converted intohyperedges for storage, then temporal characteristics are preserved, but data processing complexity increases
Solution Approach 1:
The patent extracts temporal characteristics from events and stores them as separate metadata fields (occurrence time, time window, sequence number) alongside the hyperedge structure. This extraction approach preserves complete temporal information while allowing the core hypergraph processing to remain relatively simple, as temporal attributes can be queried and analyzed independently without reprocessing the entire hypergraph structure.
Data Source
AI summary
Embodiments of this specification relate to a hypergraph data storage method and apparatus with a temporal characteristic and a hypergraph data query method and apparatus with a temporal characteristic. The storage method includes: obtaining raw edge data of a first hyperedge; storing first target data of the N nodes in a second storage table in a first file, and determining first storage location information of the N nodes in the second storage table; forming first hyperedge data based on the first number and the first storage location information, and storing the first hyperedge data in a first storage table in the first file; and storing the first file in a first file directory in a file system based on the first time, where the first file directory corresponds to a first time window, and the first time falls within the first time window.