Spatial Data Indexing Using Least Differential Sum and Binary Decision Tree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current indexing methods in database management systems, such as R-Tree, Quad-Tree, and k-d Tree, are inefficient for managing large and deviated data sets, particularly in distributed environments, as they either require excessive rebalancing or slow down during data retrieval.
Innovation Solution
The method involves dividing data into an N×M matrix with equal cells, grouping cells using the least differential sum (LDS) method, constructing a decision tree, labeling memory blocs, and linking them with resolution nets to efficiently manage and retrieve data points across multiple computers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If R-Tree is used to manage spatial data, then search speed is improved due to balanced tree structure, but data insertion efficiency deteriorates due to excessive rebalancing operations
Solution Approach 1:
The patent segments the spatial data space into an N×M matrix grid structure, where each cell can independently store data points. This segmentation eliminates the need for global rebalancing operations in R-Tree, as each matrix cell operates independently. Data insertion simply adds points to appropriate matrix cells without triggering tree restructuring, thus maintaining high insertion efficiency while preserving fast search capabilities through the grid-based organization.
2Productivity
If Quad-Tree is used to manage spatial data, then data insertion efficiency is improved due to automatic node splitting, but search speed deteriorates especially with deviated data distributions
Solution Approach 1:
The patent applies local quality by allowing different matrix cells to have different data densities and characteristics. Each cell is optimized locally based on its actual data distribution rather than forcing a uniform tree structure. This local optimization enables efficient handling of deviated data distributions where hotspots and sparse regions coexist, improving search performance while maintaining insertion efficiency through the flexible matrix structure.
3Volume of stationary object
If data is distributed across multiple computers in a network, then storage capacity is improved, but data retrieval time increases due to network latency
Solution Approach 1:
The patent introduces a hierarchical dimension to distributed data storage by organizing data across multiple computers in a multi-level matrix structure. Local matrix cells can be cached in memory while maintaining the global N×M structure across the network. This dimensional organization allows the system to exploit both local memory speed and distributed storage capacity, reducing network latency by keeping frequently accessed data locally while maintaining access to the full distributed dataset.
Data Source
AI summary
A method and a computer software for managing a database of a computer network are disclosed which comprises: selecting a data sample, dividing the data sample into an N×M matrix having equal cells, grouping adjacent cells into a plurality of memory blocs by a least differential sum method, constructing a decision tree based on the partitions of the memory blocs, labeling each of the memory blocs based on the decision tree; storing and retrieving data points into the memory blocs; and linking the memory blocs together by a set of resolution nets.