Selective Indexing for Multi-Hop Graph Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Complex querying of online professional network data is time- and resource-intensive due to the exponential increase in the number of links to be scanned with the number of hops in the query, particularly exacerbated by nodes with a large number of connections, leading to significant computational resources and time consumption.
Innovation Solution
The implementation of a selective indexing apparatus that identifies frequently used attributes and nodes with large numbers of edges to create targeted indexes, reducing the search space and processing requirements for complex queries, thereby mitigating latency and memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If complex queries scan all links between individuals and organizations to find paths, then query completeness is improved, but query execution time increases exponentially with the number of hops
Solution Approach 1:
The patent segments the network graph by identifying and separating high-degree nodes (nodes with many connections) from the rest of the graph. By creating a segmented structure where high-degree nodes are explicitly identified and handled differently, the system can process queries more efficiently by avoiding exhaustive scanning of all links while still ensuring query completeness through targeted processing of segmented portions.
Solution Approach 2:
The patent performs preliminary action by pre-identifying and indexing high-degree nodes before query execution. This preliminary classification of nodes based on their degree allows the query processing mechanism to leverage this pre-computed information, avoiding the need to scan all links during query execution and thereby reducing execution time while maintaining completeness.
2Measurement precision
If queries include nodes with large numbers of connections in the search space, then query accuracy is improved, but computational resource consumption increases significantly
Solution Approach 1:
The patent applies local quality by treating high-degree nodes differently from regular nodes in the graph. Instead of applying a uniform processing approach to all nodes, the system identifies nodes with locally high connection density and applies specialized handling (such as pre-computed indexes or cached results) specifically to these local regions, thereby reducing overall computational resource consumption while maintaining query accuracy.
Solution Approach 2:
The patent changes the parameter of node representation by introducing a degree-based classification parameter. Nodes are transformed from a uniform structure into differentiated categories based on their connection degree, allowing the query processing to adapt its resource usage based on the specific parameters of the nodes involved, thus reducing computational overhead for high-degree nodes while preserving accuracy.
3Adaptability or versatility
If the number of hops in the query path increases, then query comprehensiveness is improved, but the number of links to scan increases exponentially
Solution Approach 1:
The patent implements a nested structure by organizing the graph processing into hierarchical levels. High-degree nodes serve as nested anchors that contain or reference pre-computed information about their connection patterns. This nested organization allows multi-hop queries to leverage the nested structure at each level, reducing the exponential growth of links to scan while maintaining comprehensive query results.
Data Source
AI summary
The disclosed embodiments provide a system for processing data. During operation, the system obtains a query containing two or more hops in a graph of online professional network data and one or more predicates containing one or more attributes of the online professional network data. The graph includes nodes representing members and companies, and a set of edges representing relationships between pairs of nodes. The query also includes two or more input nodes, including a first input member and/or an input company, that are used to form a path containing the two or more hops, along with an output that matches the one or more predicates and includes one or more output members. Next, the system executes the query to obtain a subset of the online professional network data matching the query. Finally, the system provides the subset of the online professional network data in response to the query.