We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Motif-Based Community Detection: A Probabilistic Model Based on Repeating Patterns.
- Authors
Hajibabaei, H.; Seydi, V.; Koochari, A.
- Abstract
Background and Objectives: The detection of community in networks is an important tool for revealing hidden data in network analysis. One of the signs that the community exists in the network is the neighborhood density between nodes. Also, the existence of a concept called a motif indicates that a community with a high edge density has a correlation between nodes that goes beyond their close neighbors. Motifs are repetitive edge patterns that are frequently seen in the network. Methods: By estimating the triangular motif in the network, our proposed probabilistic motif-based community detection model (PMCD) helps to find the communities in the network. The idea of the proposed model is network analysis based on structural density between nodes and detecting communities by estimating motifs using probabilistic methods. Results: The suggested model's output is the strength of each node's affiliation to the communities and detecting overlaps in communities. To evaluate the performance and accuracy of the proposed method, experiments are done on real-world and synthetic networks. The findings show that, compared to other algorithms, the proposed method is acting more accurately and densely in detecting communities. Conclusion: The advantage of PMCD in using the probabilistic generative model is speeding up the computation of the hidden parameters and establishing the community based on the likelihood of triangular motifs. In fact, the proposed method proves there is a probabilistic correlation between the observation of two node pairs in different communities and the increased existence of motif structure in the network.
- Subjects
NEIGHBORHOODS; COMMUNITIES; QUANTUM computing; STIFF computation (Differential equations); PROBABILISTIC automata
- Publication
Journal of Electrical & Computer Engineering Innovations (JECEI), 2024, Vol 12, Issue 1, p247
- ISSN
2322-3952
- Publication type
Article
- DOI
10.22061/jecei.2023.9931.663