Issue |
EPL
Volume 146, Number 5, June 2024
|
|
---|---|---|
Article Number | 51001 | |
Number of page(s) | 7 | |
Section | Statistical physics and networks | |
DOI | https://doi.org/10.1209/0295-5075/ad41be | |
Published online | 03 June 2024 |
The equivalences of community detection methods for bipartite networks
1 College of Science, Guilin University of Aerospace Technology - Guilin, Guangxi 541004, China
2 The Center for Applied Mathematics of Guangxi (Guangxi Normal University) - Guilin, Guangxi 541004, China
3 Business School, University of Shanghai for Science and Technology - Shanghai 200093, China
Received: 4 February 2024
Accepted: 22 April 2024
Investigating the community structures of bipartite networks is a frequent topic of discussion in the ecological and social fields. The most widely used methods, as proposed by numerous academics from varying perspectives, include spectral graph partitioning, modularity, non-negative matrix factorization, and stochastic block model. In this paper, we demonstrate three equivalences among these four methods. i) Both Dhillon spectral graph partitioning and Barber modularity clustering are equivalent to solving for the matrix's left and right singular vectors after relaxing the discrete constraints. ii) The non-negative matrix factorization clustering is equivalent to the Dhillon spectral graph partitioning. iii) The bipartite stochastic block model is equivalent to the constraint-based NMF that uses K-L divergence as its cost function. These equivalences, obtained through rigorous mathematical derivations, will aid in the future development of efficient algorithms for community detection in bipartite networks.
© 2024 EPLA
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.