Ad-hoc Graph Cube Querying for Dynamic OLAP Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing OLAP approaches for graph data require upfront modeling of facts and dimensions, making them inflexible to adapt to changing data volumes and business intelligence requirements, especially in scenarios involving irregularly structured data.
Innovation Solution
The implementation of a property graph model that allows for ad-hoc multidimensional analysis by defining a graph cube at the time of query, using summarization rules and dimension specifications to dynamically link facts, dimensions, and measures, enabling flexible querying without pre-defined structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If upfront modeling of facts and dimensions is performed, then analytical queries can be executed, but the system becomes inflexible to changing data volumes and business intelligence requirements
Solution Approach 1:
The patent applies dynamics by transitioning from static upfront modeling to dynamic ad-hoc model creation. The graph cube is constructed at query time rather than pre-defined, allowing the analytical model to adapt dynamically to changing data volumes and business intelligence requirements. This enables the system to remain flexible while handling evolving analytical needs without requiring complex upfront modeling for all possible scenarios.
Solution Approach 2:
The patent applies preliminary action by pre-computing and storing graph data in a property graph model with organized vertices, edges, and properties before query execution. This preliminary structuring of data allows for efficient ad-hoc analysis later without requiring complex processing during query time, thus enabling flexibility without sacrificing performance.
2Adaptability or versatility
If ad-hoc query definition is implemented, then flexibility to change analytical requirements is improved, but query processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the graph data into organized components (vertices, edges, properties) with defined relationships. This segmentation allows the query processor to efficiently handle ad-hoc queries by working with structured segments rather than unstructured data, thus reducing processing complexity while maintaining query flexibility.
Solution Approach 2:
The patent applies universality through the graph cube structure that can serve multiple analytical purposes. The same graph cube can be queried with different dimension specifications and aggregation functions to address various business intelligence requirements, reducing the need for multiple specialized processing paths and simplifying overall query processing complexity.
3Adaptability or versatility
If graph cube is defined at query time, then no pre-defined structures are needed, but computational overhead increases
Solution Approach 1:
The patent applies preliminary action by pre-computing the graph cube structure from the property graph model before query execution. This includes pre-establishing the dimensional structure, facts, and relationships in the graph cube, so that when ad-hoc queries are executed, the system only needs to apply specific dimension specifications and aggregation functions rather than building the entire structure from scratch, thus reducing query execution time while maintaining flexibility.
Data Source
AI summary
Disclosed herein are system, method, and computer program product embodiments for performing ad-hoc analytical queries of graph data. An embodiment operates by receiving a graph pattern for a subgraph of interest. The facts of interest are then selected from graph data based on the received graph pattern. Dimensions are then defined based on a dimension seed pattern and a set of level expressions, and measures are defined based on a computation function and an aggregation function. A graph cube is formed based on the selected facts and the defined dimensions and measures. Because the facts, dimensions, and measures of interest are defined at the time of an analytical query, a user does not have to define such facts, dimensions, and measures, or know which analytical queries will be of interest, at the time of data collection.