Community Detection and Visualization in Complex Network by the Density-Canopy-Kmeans Algorithm and MDS Embedding

With the increasing availability of social networks and biological networks, detecting network community structure has become more and more important. However, most traditional methods for detecting community structure have limitations in dimension reduction or parameter optimization. In this paper,...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE access Ročník 7; s. 1
Hlavní autoři: Li, Manzhi, Wang, Hongtao, Long, Haixia, Xiang, Ju, Wang, Bo, Xu, Junlin, Yang, Jialiang
Médium: Journal Article
Jazyk:angličtina
Vydáno: Piscataway IEEE 01.01.2019
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
Algorithms
Canopies
Clustering
Clustering algorithms
Community detection
Complex networks
Density
Dimensionality reduction
Heuristic algorithms
MDS
Measurement
Network
Network analysis
Optimization
Partitioning algorithms
Reduction
Social networks
the Density-Canopy-Kmeans clustering algorithm
Visualization
ISSN:2169-3536, 2169-3536
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:With the increasing availability of social networks and biological networks, detecting network community structure has become more and more important. However, most traditional methods for detecting community structure have limitations in dimension reduction or parameter optimization. In this paper, we propose a Density-Canopy-Kmeans clustering algorithm (DCK) to detect network community structure. Specifically, we define a novel distance metric, which integrates random distance and community structure coefficient based on the Jaccard distance. After applying the Multidimensional Scaling (MDS) dimension reduction, we cluster the nodes. KMEANS is combined with density clustering and canopy clustering to determine the optimal number of communities and the best initial seeds are determined to improve the accuracy and stability of the K-means algorithm. Compared with traditional community detection methods, our method has a higher classification accuracy and a better visualization effect. Thus, this method is effective for analyzing network communities.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2019.2936248