Improving the stability of label propagation algorithm by propagating from low-significance nodes for community detection in social networks

Community detection in complex network analysis is a quite challenging problem spanning many applications in various disciplines such as biology, physics and social network. A large number of methods have been developed for this problem, among which the label propagation algorithm (LPA) has attracted much attention because of its advantages of nearly-linear running time and easy implementation. Nevertheless, the random updating order and tie-breaking strategy in LPA make the algorithm unstable and may even lead to the formation of a monster community. In this paper, an improved LPA called LPA-INTIM is proposed for solving the community detection problem. Firstly, an intimacy matrix is constructed using local topology information for measuring the intimacy between nodes. And then, the node importance is calculated to ensure that nodes are updated in a specific order. Finally, the label influence is evaluated for updating node label during the label propagation process. In addition, we introduce a novel tightness function to improve the stability of the proposed algorithm. By the comparison with the methods presented in the literatures, experimental results on real-world and synthetic networks show the efficiency and effectiveness of our proposed algorithm.

Download Full-text

LPA-CBD an improved label propagation algorithm based on community belonging degree for community detection

International Journal of Modern Physics C ◽

10.1142/s0129183118500110 ◽

2018 ◽

Vol 29 (02) ◽

pp. 1850011 ◽

Cited By ~ 5

Author(s):

Chun Gui ◽

Ruisheng Zhang ◽

Zhili Zhao ◽

Jiaxuan Wei ◽

Rongjing Hu

Keyword(s):

Community Detection ◽

Maximum Degree ◽

Label Propagation ◽

General Process ◽

Network Community ◽

Propagation Algorithm ◽

Community Belonging ◽

Original Label ◽

The Stability ◽

Detection Quality

In order to deal with stochasticity in center node selection and instability in community detection of label propagation algorithm, this paper proposes an improved label propagation algorithm named label propagation algorithm based on community belonging degree (LPA-CBD) that employs community belonging degree to determine the number and the center of community. The general process of LPA-CBD is that the initial community is identified by the nodes with the maximum degree, and then it is optimized or expanded by community belonging degree. After getting the rough structure of network community, the remaining nodes are labeled by using label propagation algorithm. The experimental results on 10 real-world networks and three synthetic networks show that LPA-CBD achieves reasonable community number, better algorithm accuracy and higher modularity compared with other four prominent algorithms. Moreover, the proposed algorithm not only has lower algorithm complexity and higher community detection quality, but also improves the stability of the original label propagation algorithm.

Download Full-text

Label propagation algorithm based on edge clustering coefficient for community detection in complex networks

International Journal of Modern Physics B ◽

10.1142/s0217979214502166 ◽

2014 ◽

Vol 28 (30) ◽

pp. 1450216 ◽

Cited By ~ 10

Author(s):

Xian-Kun Zhang ◽

Xue Tian ◽

Ya-Nan Li ◽

Chen Song

Keyword(s):

Complex Networks ◽

Community Detection ◽

Learning Algorithm ◽

Low Complexity ◽

Label Propagation ◽

Clustering Coefficient ◽

Propagation Algorithm ◽

The Stability ◽

Stability And Accuracy ◽

Edge Clustering Coefficient

The label propagation algorithm (LPA) is a graph-based semi-supervised learning algorithm, which can predict the information of unlabeled nodes by a few of labeled nodes. It is a community detection method in the field of complex networks. This algorithm is easy to implement with low complexity and the effect is remarkable. It is widely applied in various fields. However, the randomness of the label propagation leads to the poor robustness of the algorithm, and the classification result is unstable. This paper proposes a LPA based on edge clustering coefficient. The node in the network selects a neighbor node whose edge clustering coefficient is the highest to update the label of node rather than a random neighbor node, so that we can effectively restrain the random spread of the label. The experimental results show that the LPA based on edge clustering coefficient has made improvement in the stability and accuracy of the algorithm.

Download Full-text