Social Network Community Discovery via Label Propagation Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing community discovery methods, such as the Scalable Community Detection (SCD) algorithm, often result in overly fine partitions of social networks, failing to meet user requirements for coarser granularity, such as grouping classmates or colleagues, due to their high accuracy leading to smaller community sizes.
Innovation Solution
A community discovery method that partitions nodes in a social network into initial communities using a label propagation algorithm, updating community labels based on the quantity of neighboring nodes to merge communities, thereby obtaining coarser and more accurate community structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the SCD algorithm partitions the social network into communities based on triangles, then the accuracy of community detection is improved, but the granularity becomes too fine (communities become too small)
Solution Approach 1:
The patent merges multiple fine-grained communities obtained from triangle-based partitioning into coarser communities. Specifically, it combines communities sharing the same triangle-based label into larger communities, thereby increasing community size while preserving detection accuracy. This resolves the contradiction by combining the benefits of accurate fine-grained detection with the need for larger, more meaningful communities.
2Reliability
If the SCD algorithm performs fine partitioning, then connections inside communities are closed (high accuracy), but the partition does not satisfy user requirements for coarser granularity
Solution Approach 1:
The patent introduces a dynamic two-stage partitioning process that adapts to different user needs. The first stage performs fine-grained partitioning for accuracy, while the second stage dynamically merges communities based on user-specified granularity requirements. This dynamic approach allows the system to provide both fine-grained and coarse-grained partitions as needed, resolving the contradiction between reliability and adaptability.
3Measurement precision
If the SCD algorithm partitions based on triangles, then community detection accuracy is high, but the method is not applicable to many scenarios requiring coarser partitions
Solution Approach 1:
The patent creates a universal community detection framework that can handle multiple scenarios by combining triangle-based partitioning with a merging mechanism. The system can output communities at different granularity levels depending on the scenario requirements, making it universally applicable to both fine-grained and coarse-grained needs. This resolves the contradiction by making the method versatile across different application scenarios.
Data Source
AI summary
A community discovery method is provided. The community discover method includes partitioning nodes in a social network into community nodes partitioned into n first communities, each of the n first communities being associated with a corresponding community label, the corresponding label of each of the community nodes initially indicating a first community from among the n first communities to which the community node belongs, and n being an integer greater than or equal to 2; updating the corresponding label of each community node comprised in the n first communities; and partitioning the community nodes into m second communities, each of the community nodes in each of the m second communities having a same label, and m being a positive integer less than n.