Neural Query Graph Execution for Database Dependency Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing database management systems face inefficiencies in query processing due to the lack of effective methods for optimizing the execution of queries in non-volatile memory systems, particularly in relational databases, where the dependency relationships between processes are not adequately addressed.
Innovation Solution
A database management system utilizing a directed graph with nodes representing processes and edges representing input and output, combined with neural networks to calculate features of nodes, where the order-dependent AND nodes and order-invariant OR nodes are processed differently, optimizing query execution paths based on these features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional tree structure is used for query processing optimization, then the system structure is simple, but the query processing efficiency is insufficient
Solution Approach 1:
The patent segments the query processing into distinct node types (AND nodes and OR nodes) within a directed acyclic graph structure. Each node type is handled by specialized neural networks that process them differently, allowing for optimized processing of each segment while maintaining an overall structured approach to query optimization.
Solution Approach 2:
The patent transitions from a traditional tree structure to a directed acyclic graph structure, adding dimensional complexity to represent query processing more accurately. This dimensional change allows multiple parent nodes and more flexible representation of query dependencies, improving processing efficiency despite increased structural complexity.
2Measurement precision
If uniform neural network is used for all nodes, then the system is easy to implement, but the processing accuracy for different node types is insufficient
Solution Approach 1:
The patent applies local quality by using different neural network configurations for different node types. AND nodes use order-dependent neural networks that consider input sequence, while OR nodes use order-invariant neural networks that process inputs regardless of sequence. This localized specialization improves feature calculation accuracy for each node type while maintaining overall system coherence.
3Productivity
If dependency relationships between processes are not addressed, then the system is simple to manage, but the query execution optimization is insufficient
Solution Approach 1:
The patent implements feedback mechanisms where the neural networks calculate features of nodes based on their input dependencies, and these features feed back into the query execution plan generation. The system continuously refines the execution plan by considering the calculated features and their relationships, optimizing query execution through iterative feedback loops.
Solution Approach 2:
The patent introduces feature calculations as intermediaries between the directed graph structure and the query execution optimization. These features act as mediators that capture the essence of node dependencies and relationships, enabling the system to optimize query execution without directly managing complex dependency relationships in a cumbersome way.
Data Source
AI summary
A database management system includes a processor and a memory, wherein the processor is configured to calculate features of each of a first node and a second node of a directed graph, calculate the feature of the first node by inputting the feature of the second node to be input to the first node to a first neural network that is dependent on an input order of the first node, and calculate the feature of the second node by inputting the feature of the first node to be input to the second node to a second neural network that is not dependent on an input order of the second node.