Graph Database Island Detection via Zone-Based Designators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for searching islands of nodes within graph databases are inefficient, especially as graphs become larger and more dense, due to the high computational complexity involved.
Innovation Solution
The method involves subdividing graphs into zones and subzones by assigning designators with dimensional values that indicate node membership, allowing for efficient identification of islands by checking connections between nodes with matching patterns of dimensional values, thereby reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If prior art algorithms are used to search for islands in graph databases, then the search can be performed, but the computational complexity becomes prohibitively high as the graph grows larger and denser
Solution Approach 1:
The patent divides the graph database into multiple zones based on spatial or structural characteristics. Each zone contains a subset of nodes and edges, allowing the island search algorithm to operate on smaller, manageable segments rather than the entire graph. This segmentation reduces the computational complexity from O(V+E) for the whole graph to O(v_i + e_i) for each zone i, where v_i and e_i are the nodes and edges in zone i.
Solution Approach 2:
The patent introduces a spatial or structural dimension to organize graph nodes into zones, transforming the traditional single-dimensional graph representation into a multi-dimensional structure. This dimensional change enables the use of spatial indexing or zone-based filtering techniques to quickly eliminate regions that cannot contain islands, reducing the search space without losing completeness.
2Quantity of substance
If the graph database becomes larger and more dense, then more information can be stored, but the time required to search for islands increases significantly
Solution Approach 1:
The patent performs preliminary actions by pre-computing and storing zone assignments for each node, and pre-identifying boundary nodes that connect different zones. This preliminary organization allows the island search algorithm to quickly jump between zones or eliminate entire zones from consideration, rather than examining every node and edge sequentially. The pre-computed zone structure acts as an index that accelerates the search process.
Solution Approach 2:
By segmenting the large dense graph into multiple zones, the patent enables parallel processing of island searches across different zones. Multiple processors can simultaneously search for islands in different zones, reducing the overall search time from linear O(V+E) to approximately O((V+E)/p) where p is the number of processors, while maintaining the ability to handle arbitrarily large graphs.
Data Source
AI summary
The invented method includes one or more of the following aspects of selecting a multiplicity of nodes, deriving a designator for each node of the multiplicity of nodes and associating each designator in a one-to-one correspondence with a respective node from which a respective designator is derived, each designator indicating a plurality of dimensional values, and each dimensional value expressing a quality of a variate property of an associated node, selecting a unique designator having a particular pattern of dimensional values, selecting a first node having a first node designator matching the particular pattern of dimensional values, and when no node is found to have the particular pattern of dimensional values expressed in the respective designator to be connected by an edge to an external node, reporting a positive finding of an island.