Denormalized Virtual Table for Semi-Structured Data Querying

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Querying large volumes of semi-structured data, such as JSON and XML, is inefficient due to the need to traverse complex hierarchies, especially when schema information is unknown, leading to costly operations and suboptimal performance even after normalization or denormalization.

Innovation Solution

A denormalized data arrangement is generated by flattening semi-structured data into a flat relational scheme, allowing queries to extract data directly from columns using predicates and scan or join techniques, with path expressions converted into relational statements for efficient execution.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Stability of the object's composition

If semi-structured data is stored in hierarchical documents to preserve hierarchical information, then data storage convenience and hierarchical information preservation are improved, but query performance deteriorates due to costly hierarchical traversal operations

Engineering Contradiction:
Improvehierarchical information preservationVSAvoidquery performance
Core Design Contradiction:
Stability of the object's compositionVSProductivity

Solution Approach 1:

The patent segments the hierarchical semi-structured data into multiple normalized relational tables based on the hierarchy levels and relationships. Each table represents a specific level or aspect of the hierarchy, allowing queries to access only the necessary segments without traversing the entire hierarchical structure, thus improving query performance while preserving hierarchical information through foreign key relationships.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary mapping layer that translates hierarchical data access patterns into efficient relational database operations. This intermediary mechanism handles the conversion between hierarchical navigation and relational table joins, eliminating the need for expensive hierarchical traversal while maintaining the logical hierarchical structure through relational integrity constraints.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Quantity of substance

If semi-structured data is normalized and stored in separate relational tables to avoid repetition, then storage efficiency is improved, but query performance deteriorates due to substantial join operations consuming resources

Engineering Contradiction:
Improvestorage efficiencyVSAvoidquery performance
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The patent performs preliminary action by pre-computing and materializing common join results into denormalized views or summary tables. These pre-computed structures cache the results of frequent join operations, allowing queries to retrieve data from these optimized structures without performing expensive real-time joins, thus improving query performance while maintaining storage efficiency through selective denormalization.

Inventive Principle:
Principle #10Preliminary action

3Device complexity

If data is stored in a single normalized table to simplify structure, then device complexity is reduced, but storage efficiency deteriorates due to repetition of data

Engineering Contradiction:
Improvedata structure simplicityVSAvoidstorage efficiency
Core Design Contradiction:
Device complexityVSQuantity of substance

Solution Approach 1:

The patent segments the data into multiple normalized tables organized by hierarchical levels and relationship types. This segmentation eliminates data repetition by storing each piece of data in only one table, while maintaining referential integrity through foreign keys. The segmented structure provides both storage efficiency and manageable complexity through clear separation of concerns.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3365804B1Columnar data arrangement for semi-structured data
Publication Date: 2020.08.05 ORACLE INT CORP
  • EP3365804B1 patent drawingFigure 1A
  • EP3365804B1 patent drawingFigure 1B
  • EP3365804B1 patent drawingFigure 2

AI summary

Techniques are provided for denormalizing semi-structured hierarchical data into a virtual table. At least portion of semi-structured data document collection is denormalized for improving the execution of queries that involves a traversal of the semi-structured data hierarchy of the semi-structured data document collection, in an embodiment. Based on the extracted schema of the semi-structured data, a denormalized arrangement is generated in which the hierarchical relationship of the semi-structured data is converted into a set of columns. The denormalized arrangement is materialized in a virtual table by applying the denormalized arrangement onto the semi-structured data. A received query that involves a traversal of the semi-structured data hierarchy is converted to a relational query that can be executed on the virtual table in an embodiment. The execution of the relational query on the virtual table improves the performance in generating the resulting data set.